Easy ReviewDOP-C02
SDLC Automation119

Systems Manager

Configure managed nodes, run commands and automation, patch fleets, control sessions, and store application parameters.

Este conteúdo de estudo ainda não foi traduzido, então aparece em inglês. Ajude a traduzir

Systems Manager connects managed nodes to configuration, patching and remote-access tools. Use Run Command for a document execution, State Manager for recurring configuration, and Automation for coordinated maintenance actions. Parameter Store holds named application configuration values.

Systems Manager basics

AWS Systems Manager is the operations hub for servers: it patches, runs commands, opens sessions and stores parameters for EC2 instances and on-premises machines.

How it works

  1. Install SSM Agent on the node and let it reach Systems Manager; it becomes a managed nodeA machine configured to receive Systems Manager management actions through SSM Agent..
  2. Use tools such as Run Command, Session Manager, Automation and Parameter Store against one node or many.
  3. The same console covers AWS, on-premises and other cloud nodes.

Exam cues

  • If the question says manage or patch a fleet of servers, including on premises, think Systems Manager.
  • If the question says shell access without SSH or bastionA separate host used to reach machines in a private network. hosts, think Session Manager.

Remember

  • Needs SSM Agent on each node.
  • Works for AWS, on-premises and multicloud nodes.

Trap: Systems Manager is not only for EC2: on-premises servers become managed nodes too.

Da documentação da AWS (4)
  • AWS Systems Manager centrally views, manages and operates nodes at scale in AWS, on premises and in other clouds.
  • A node is managed by Systems Manager when SSM Agent is installed on it and can reach the service.
  • Systems Manager manages nodes remotely without bastion hosts, SSH or remote PowerShell, and its tools include Run Command, Session Manager, Automation and Parameter Store.
  • Systems Manager gives a centralized view to see and manage EC2 instances, on-premises servers and other managed nodes across an organization's accounts.

Managed nodes and hybrid activation

A Systems Manager managed nodeA machine configured to receive Systems Manager management actions through SSM Agent. is a machine configured to receive management actions through SSM Agent. The agent is the software on the machine that connects it to Systems Manager.

How it works

  1. Install and run SSM Agent. IAM is the AWS permissions system. For EC2, grant node permissions through an instance profileA container that passes an IAM role to an EC2 instance., the container for an IAM role attached to an instance, or through Default Host Management Configuration (DHMC). The instance role trusts ec2.amazonaws.com; the hybrid service roleAn IAM role that an AWS service assumes to perform authorized actions. trusts ssm.amazonaws.com.
  2. Give the agent an outbound HTTPS path. An interface VPC endpointA private network entry point for connecting to a supported service through AWS PrivateLink. is a private network entry point to an AWS service. For private access, allow port 443 from the nodes into the endpoint security groupA set of rules that controls permitted network traffic for associated resources., the rules that control its network traffic.
  3. For non-EC2 servers, call CreateActivation, choose a service role that trusts ssm.amazonaws.com, and register SSM Agent with the returned activation ID and code. The resulting node ID starts with mi-.
  4. For Greengrass core devices, configure the registration service role and token exchange roleAn IAM role that supplies temporary AWS permissions to Greengrass device components., which supplies temporary AWS permissions to device components. Deploy the SSM Agent component; Greengrass performs registration.

Exam cues

  • If the question says an EC2 fleet needs node permissions without per-instance profiles, think DHMC, because it supplies permissions within the enabled account and Region.
  • If the question says a private node is missing, think agent, IAM and endpoint checks, because the agent needs all three to connect.
  • If the question says Greengrass core devices need management, think the Greengrass SSM Agent component, because its registration flow includes the token exchange role.

Remember

  • DHMC requires IMDSv2The token-based interface through which software on an EC2 instance retrieves instance metadata., the token-based EC2 metadataDescriptive information such as installed software versions or file attributes rather than file contents. service, and SSM Agent 3.2.582.0 or later.
  • Use ssm and ssmmessages; agent 3.3.40.0 and later prefer ssmmessages. Regions launched in 2024 or later do not support ec2messages.
  • Allow required S3 access for agent updates, scripts, output and patching. Private logging and KMS integrations also need their endpoints.
  • Activation defaults: 24-hour expiration and one registration. Maximums: expiration 30 days ahead and 1000 registrations per activation.
  • June 2026: the hybrid advanced-instances tier was removed. September 2026: hybrid Session Manager and Run Command moved to pay-as-you-go pricing.

Trap: Enabling DHMC does not override an instance profile that permits ssm:UpdateInstanceInformation. That profile takes precedence.

Trap: The 1000-registration limit is per activation, not a fleet limit.

Agent connections originate on the managed node
12Managed nodeSSM AgentVPC endpointsssm and ssmmessagesSystems Managermanagement service
  1. SSM Agent initiates an outbound HTTPS connection to a private endpoint on port 443.

  2. The endpoint provides private access to Systems Manager; node permissions still authorize the calls.

Da documentação da AWS (32)
  • Managed nodes run SSM Agent to communicate with Systems Manager.
  • EC2 node permissions can come from an IAM instance profile or Default Host Management Configuration.
  • An instance profile passes an IAM role to an EC2 instance.
  • AmazonSSMManagedInstanceCore grants core Systems Manager node permissions.
  • SSM Agent initiates connections to Systems Manager, so nodes need outbound HTTPS on port 443 and no inbound Systems Manager port.
  • Private access uses interface VPC endpoints; their security groups must accept port 443 from the nodes.
  • Private Systems Manager access uses ssm and ssmmessages endpoints.
  • Private S3 access supports agent updates, S3 scripts, output and patch operation buckets.
  • Private CloudWatch logging requires the logs endpoint.
  • Private KMS integration can require the kms endpoint.
  • SSM Agent 3.3.40.0 and later prefer ssmmessages when available.
  • Regions launched in 2024 or later do not support ec2messages.
  • An attached instance profile allowing ssm:UpdateInstanceInformation takes precedence over Default Host Management Configuration credentials.
  • CreateActivation returns an activation ID and code for registering non-EC2 machines with SSM Agent.
  • A hybrid activation assigns a service role that trusts ssm.amazonaws.com, rather than an EC2 instance profile.
  • Hybrid-activated machines receive managed node IDs beginning with mi-.
  • An activation expires after 24 hours by default.
  • An activation expiration can be set up to 30 days ahead.
  • CreateActivation defaults to one registration.
  • CreateActivation permits a registration limit up to 1000 per activation.
  • Greengrass core devices need a Systems Manager registration service role, an updated token exchange role and deployment of the SSM Agent component; Greengrass then registers them automatically.
  • The hybrid advanced-instances tier was removed in June 2026.
  • Session Manager and Run Command on hybrid nodes use pay-as-you-go pricing starting September 2026.
  • The IAM role attached through an EC2 instance profile trusts ec2.amazonaws.com.
  • IMDSv2 uses session tokens when software retrieves EC2 instance metadata.
  • The Greengrass token exchange role grants device components temporary AWS permissions.
  • IAM: The AWS permissions system that controls which identities can perform actions on resources.
  • Instance profile: A container that passes an IAM role to an EC2 instance.
  • Interface VPC endpoint: A private network entry point for connecting to a supported service through AWS PrivateLink.
  • A set of rules that controls permitted network traffic for associated resources.
  • The Config rule ec2-instance-managed-by-systems-manager marks EC2 instances NON_COMPLIANT when Systems Manager doesn't manage them, and DescribeInstanceInformation lists the managed nodes.
  • Systems Manager Fleet Manager is a console view for remotely managing nodes on AWS or on premises, showing fleet health and running common troubleshooting tasks.

Run Command and remote scripts

Run Command executes a document on managed nodes so an operator can apply the same action across a fleet. A document is a reusable recipe of steps with inputs supplied when it runs.

How it works

  1. Choose a JSON or YAML command document and its version. Supply the parameters for that execution.
  2. Select node IDs, tagsA key and optional value attached to a resource to identify or categorize it. or a resource groupA saved collection of resources selected by a query such as resource types and tags.. A tag-based resource group is a saved selection by resource type and tag filtersA resource-selection condition matching a tag key and optionally its values.. Resource groups are maintained separately in each Region.
  3. For remote scripts, use AWS-RunRemoteScript. For a downloaded SSM documentA reusable JSON or YAML definition of actions, steps and input parameters for Systems Manager., use aws:downloadContent followed by aws:runDocument with documentType: LocalPath. A downloaded Linux shell script uses aws:runShellScript.
  4. Retain complete output in S3 or CloudWatch Logs. Configure SNS notifications for the overall command or for each invocationA copy of a Run Command command sent to a particular managed node., the copy of the command sent to a particular node.

Exam cues

  • If the question says scripts already live in GitHub or S3, think AWS-RunRemoteScript, because it downloads and runs them.
  • If the question says a reviewed document revision must be repeatable, think an explicit document versionA saved revision of an SSM document that can be selected explicitly for execution., because editing content creates a new revision.
  • If the question says Auto Scaling launch events must configure managed nodes, think EventBridge targeting Run Command, because Run Command is a supported event target.
  • If the question says react to an alarm by running a command on an instance, think SNS to Lambda to Run Command SendCommand, because the instance itself cannot subscribe to the topic.

Remember

  • Tag targets support five keys and five values per key.
  • The default command response contains only the first 24000 output characters. Use S3 or CloudWatch Logs for complete output.
  • An EC2 node downloading S3 scripts needs s3:ListBucket and s3:GetObject.
  • GitHub downloads use sourceType: GitHub, repositoryA source location from which an operating-system package manager obtains updates. sourceInfo, and getOptions to select a branch or commit.
  • SNS FIFO topics are unsupported for these status notifications.

Trap: aws:downloadContent only downloads content. A later execution step must run the downloaded script or document.

Run Command sends instructions; the node retrieves the script
12Run Commandcommand documentManaged nodedownload and executeS3 bucketremote script
  1. Run Command delivers the document to SSM Agent on the targeted managed node.

  2. The node retrieves the S3 script using its node permissions and runs the requested script.

Da documentação da AWS (17)
  • SSM documents define parameterized actions and steps in JSON or YAML.
  • Editing document content increments its version; callers can choose a specific version and owners can change the default version.
  • Run Command runs command documents on managed nodes targeted by IDs, tags or a resource group.
  • Run Command tag targeting supports up to five keys and five values per key.
  • A tag-based resource group selects resources using resource types and tag key-value filters.
  • AWS-RunRemoteScript downloads and runs remote scripts from GitHub or S3.
  • For S3 scripts, the EC2 instance profile needs s3:ListBucket and s3:GetObject.
  • aws:downloadContent downloads scripts or documents; GitHub sourceInfo identifies the owner, repository and path, with getOptions selecting a branch or commit.
  • aws:runDocument executes an SSM document downloaded to a local path; aws:runShellScript executes a Linux shell script.
  • Run Command returns only the first 24000 output characters by default; S3 or CloudWatch Logs can receive complete output.
  • Run Command SNS notifications can report overall command status or individual node invocations.
  • SNS FIFO topics are unsupported for Run Command status notifications.
  • EventBridge can target Run Command to configure newly launched Auto Scaling nodes or run patching after noncompliance events.
  • Resource groups are maintained separately for each AWS Region.
  • Resource group: A saved collection of resources selected by a query such as resource types and tags.
  • Invocation: A copy of a Run Command command sent to a particular managed node.
  • The SendCommand API runs an SSM document on managed nodes chosen by instance IDs or tags, so code such as a Lambda function can trigger Run Command.

Rate controls and error thresholds

Run Command rateA schedule expression specifying a repeating time interval. controls limit how much of a fleet runs a command at once and when errors stop further dispatch. ConcurrencyThe number of target invocations allowed to run at the same time. means the number of node invocationsA copy of a Run Command command sent to a particular managed node. running at the same time.

How it works

  1. Set max-concurrency as a node count or percentage of targets. Dispatch starts with one node and ramps up as nodes acknowledge.
  2. Set max-errors as the allowed error count or percentage. Run Command stops sending new invocations when the threshold is exceeded.
  3. Allow running invocations to finish. Use concurrency of one when additional concurrent failures would be unacceptable.

Exam cues

  • If the question says limit simultaneous changes, think max-concurrency, because it controls active node invocations.
  • If the question says stop dispatch after the first error, think max-errors: 0, because zero errors are allowed.
  • If the question says avoid failures from other in-flight commands, think concurrency of one, because nodes then run serially.

Remember

  • Run Command defaults: max-concurrency: 50, max-errors: 0.
  • With max-errors: 3, dispatch stops on the fourth error.
  • Running invocations can finish and fail after new dispatch stops.

Trap: An error thresholdThe allowed error count or percentage before Run Command stops sending new invocations. controls dispatch, not an absolute failure ceiling. Even serial execution must receive the error that exceeds the allowed count before dispatch stops.

Da documentação da AWS (6)
  • Run Command max-concurrency limits simultaneous node invocations and accepts a number or percentage; the default is 50.
  • Run Command starts with one invocation and increases dispatch exponentially as nodes acknowledge, up to max-concurrency.
  • Run Command max-errors accepts a count or percentage; the default is zero, which stops new dispatch after the first error.
  • A max-errors count of three stops new dispatch when the fourth error is received.
  • Invocations already running are allowed to complete after the error threshold is exceeded and can also fail.
  • Concurrency of one makes invocations proceed one at a time to avoid additional concurrent failures after an error threshold is exceeded.

Automation runbooks and approvals

Systems Manager Automation runs a runbookAn ordered set of maintenance or deployment actions executed by Systems Manager Automation., an ordered set of maintenance or deployment actions. It can coordinate AWS resource operations and pause for a person to approve a step.

How it works

  1. Choose the runbook and its inputs. AutomationAssumeRole is the IAM role whose permissions the runbook uses; AWS-ResizeInstance uses the initiating user permissions if that role is omitted.
  2. Add aws:approve before the action that needs approval. Specify approvers and the required approval count. If sending SNS notifications, use a topic whose name starts with Automation. Approve or reject in the console or with SendAutomationSignal.
  3. Use a runbook suited to the task. AWS-UpdateLinuxAmi produces a patched Amazon Machine ImageA template used to launch an EC2 instance with its operating system and software. (AMI), the template used to launch an instance. IncludePackages and ExcludePackages control package selection; PreUpdateScript and PostUpdateScript add scripts around the update. A golden AMIAn organization-maintained base machine image used for consistent instance launches. is the organization’s maintained base image. The AWS tutorial invokes Lambda afterward to publish the new image ID in Parameter Store.
  4. For an unreachable instance, AWSSupport-ExecuteEC2Rescue stops and backs it up, attaches its root volumeThe storage volume containing the operating system used to boot an instance., the disk used to boot the operating system, to a temporary managed helper, attempts repairs, then restores the volume and restarts the original instance.

Exam cues

  • If the question says change an EC2 instance type with a reusable workflow, think AWS-ResizeInstance, because the runbook performs that change.
  • If the question says publish the new golden AMI ID, think Automation followed by a Parameter Store update, because the AWS tutorial connects those steps with Lambda.
  • If the question says Windows boot or firewall settings prevent access, think EC2Rescue, because it can inspect boot and connectivity issues through a helper.

Remember

  • Approval timeout: seven days by default, up to 30 days with timeoutSeconds.
  • MinRequiredApprovals defaults to one.
  • NotificationArn is optional; a supplied SNS topic name must start with Automation.
  • aws:approve does not support multi-account and Region automations.
  • AWSSupport-ExecuteEC2Rescue does not support encrypted root volumes.

Trap: A seven-day approval timeout is the default, not an unchangeable maximum.

Trap: An unreachable target can be repaired through a separate managed helper; the rescue runbook does not depend on running the repair command inside the unreachable operating system.

An approval step gates later runbook actions
12Automationstart runbookApproversapprove or rejectAutomationcontinue if approved
  1. The runbook pauses at aws:approve for designated principals.

  2. The runbook resumes after the required approval count is reached.

EC2Rescue repairs the root volume through a helper
12Original instancestop and back upHelper instancerepair attached volumeOriginal instancereattach and restart
  1. Automation stops and backs up the original instance, then attaches its root volume to the managed helper.

  2. After EC2Rescue attempts repair, Automation returns the root volume and restarts the original instance.

Da documentação da AWS (25)
  • Automation runbooks orchestrate maintenance and deployment actions, including calls to AWS APIs and actions on managed nodes.
  • AutomationAssumeRole supplies the IAM role used for runbook actions; without a role, AWS-ResizeInstance uses the initiating user permissions.
  • aws:approve pauses a runbook until designated principals approve or reject the step.
  • aws:approve defaults to a seven-day timeout.
  • aws:approve supports a maximum timeout of 30 days through timeoutSeconds.
  • NotificationArn is optional for aws:approve.
  • An SNS topic specified for aws:approve must have a name starting with Automation.
  • MinRequiredApprovals defaults to one.
  • MinRequiredApprovals cannot exceed the number of designated approvers.
  • aws:approve does not support multi-account and Region automations.
  • AWS-UpdateLinuxAmi updates a source AMI with Linux packages and Amazon software and creates a new AMI using a workspace instance.
  • The golden AMI tutorial patches an AMI and invokes a Lambda function at the end to update a Parameter Store value with the new AMI ID.
  • AWS-ResizeInstance changes the instance type of an EC2 instance.
  • AWSSupport-ExecuteEC2Rescue uses a temporary SSM-enabled helper to diagnose and attempt repair of an unreachable instance root volume.
  • EC2Rescue Automation stops and backs up the original instance, attaches its root volume to the helper, runs EC2Rescue, reattaches the volume and restarts the original instance.
  • AWSSupport-ExecuteEC2Rescue does not support instances with encrypted root volumes.
  • EC2Rescue for Windows can diagnose boot issues and connectivity problems caused by firewall, RDP or network interface settings.
  • SendAutomationSignal supports Approve and Reject signals for an Automation execution.
  • AWS-UpdateLinuxAmi supports IncludePackages and ExcludePackages to control package updates.
  • AWS-UpdateLinuxAmi supports PreUpdateScript and PostUpdateScript URLs for scripts before and after updates.
  • AMI: A template used to launch an EC2 instance with its operating system and software.
  • Root volume: The storage volume containing the operating system used to boot an instance.
  • An organization-maintained base machine image used for consistent instance launches.
  • An Automation runbook can run commands on instances as ordered steps with aws:runCommand and stops when a step fails.
  • Automation execution status changes are sent to EventBridge, so a rule on Failed or TimedOut can notify an SNS topic.

State Manager associations

State Manager maintains a desired configuration on managed nodes by reapplying a document. An associationA saved configuration binding a document and its inputs to targets and an execution schedule. is the saved link between that document, its inputs, the targets and the schedule.

How it works

  1. Describe the desired state in an SSM documentA reusable JSON or YAML definition of actions, steps and input parameters for Systems Manager., such as installing software or configuring an agent.
  2. Create an association with document parameters, targets and a cronA schedule expression specifying calendar times for an action. or rateA schedule expression specifying a repeating time interval. schedule. Cron names calendar times; rate expressions specify a repeating interval.
  3. By default, run at creation and again on schedule. State Manager also attempts execution when matching managed nodes first come online.
  4. Review association complianceThe reported state of a node compared with required patching or configuration criteria. and execution results. For Chef recipes, use AWS-ApplyChefRecipes on Linux, with optional compliance checks and S3 results.

Exam cues

  • If the question says reapply settings and report compliance, think State Manager, because an association keeps configuration tied to targets and a schedule.
  • If the question says new tagged nodes need an agent configuration, think a command-document association targeting those tagsA key and optional value attached to a resource to identify or categorize it., because matching nodes receive the association when first online.
  • If the question says existing Chef recipes must run on Linux, think AWS-ApplyChefRecipes, because it downloads and runs the cookbooks.

Remember

  • Associations run at creation by default. ApplyOnlyAtCronInterval can defer execution to the next specified cron interval.
  • Association cron expressions do not support months.
  • AWS-ApplyChefRecipes supports Chef 11 through 18 and Git, GitHub, HTTP or S3 sources.
  • Compliance includes Compliant or Non-compliant and an assigned severity.
  • Automation-based associations have additional requirements for automatic target updates, including creation through Quick Setup.

Trap: Recurring enforcement does not mean every manual configuration change is detected and repaired immediately. The schedule and documented triggers determine execution.

Da documentação da AWS (15)
  • A State Manager association binds an SSM document, parameters, targets and a schedule to maintain a desired configuration.
  • Associations can target node IDs, tags, resource groups or all nodes in the current account and Region.
  • By default, State Manager runs an association at creation and on its schedule.
  • State Manager attempts to apply associations when matching managed nodes come online for the first time.
  • Association schedules accept cron or rate expressions; association cron expressions do not support months.
  • ApplyOnlyAtCronInterval can prevent association execution when targets change and defer execution to the specified cron interval.
  • Systems Manager Compliance reports association Compliant or Non-compliant status and the assigned severity.
  • AWS-ApplyChefRecipes runs Chef recipes on Linux managed nodes.
  • AWS-ApplyChefRecipes supports Chef 11 through 18.
  • AWS-ApplyChefRecipes can download cookbooks from Git, GitHub, HTTP or S3.
  • AWS-ApplyChefRecipes can install the Chef client.
  • AWS-ApplyChefRecipes can optionally run compliance checks with results stored in S3.
  • Automatic target updates for Automation-based associations require a Quick Setup-created association with supported node resource types, parameters and targets.
  • Cron expression: A schedule expression specifying calendar times for an action.
  • Rate expression: A schedule expression specifying a repeating time interval.

Patch baselines and groups

A Patch Manager baselineRules and explicit lists that determine approved and rejected patches. is the rulebook that decides which patches are approved. In patch-group operations, a node tagA key and optional value attached to a resource to identify or categorize it. selects the baseline for that node and operating system.

How it works

  1. Choose an AWS predefined baseline or create a custom one. Copy a predefined baseline when its rules need changing.
  2. For patch-group operations, tag each node with Patch Group or PatchGroup and register the value to a baseline for its operating system.
  3. When patching runs, Patch Manager looks up that mapping. If none exists, it uses the current default baseline for the operating system.
  4. For custom Linux sources, specify alternative repositories, the locationsOne AWS account in one AWS Region used as an Automation execution location. that supply patch packages. Include every repositoryA source location from which an operating-system package manager obtains updates. needed for the operation, including ordinary defaults.
Approval choice Effect
Auto-approval delay Wait after patch release or update
CutoffThe time before a maintenance window ends when scheduling of new tasks stops. date Approve releases or updates on or before a fixed date
Rejected list Reject a listed patch even when also approved

Exam cues

  • If the question says Windows updates must be approved sooner than seven days, think a custom baseline, because the predefined Windows delay is seven days.
  • If the question says test and production nodes need different approved patch sets, think patch groupsNodes whose patch-group tag value is registered with an operating-system-specific patch baseline. mapped to baselines, because the node tag selects its baseline.
  • If the question says Linux updates come from a corporate repository, think custom baseline patch sources, because alternative repositories select the sources for that operation.

Remember

  • One patch group per node; one baseline per operating system per patch group.
  • Use PatchGroup when instance metadataDescriptive information such as installed software versions or file attributes rather than file contents. tags are enabled.
  • December 2022: patch-group console support became limited to account-Region pairs already using it before December 22.

Trap: Adding an alternative repository replaces the sources used for that patching operation. Include default repositories too if their updates are required.

Trap: A baseline approves patches; approval alone does not install them.

Da documentação da AWS (16)
  • Patch baselines define rules and explicit lists for approving or rejecting patches.
  • AWS predefined patch baselines cannot be customized; copying one produces an editable custom baseline.
  • Windows Server predefined baselines use a seven-day auto-approval delay.
  • Auto-approval delay waits a chosen number of days after patch release or update.
  • An approval cutoff date approves patches released or updated on or before that date.
  • A patch listed as both approved and rejected in a baseline is rejected.
  • Patch groups use the case-sensitive tag keys Patch Group or PatchGroup.
  • Nodes with instance metadata tags enabled must use PatchGroup without a space.
  • A managed node belongs to one patch group.
  • A patch group can register one baseline per operating system type.
  • If a node has no patch group or its patch group has no registered baseline, Patch Manager uses the current default baseline for the operating system.
  • Patch group functionality is unavailable in the console for account-Region pairs that did not use it before December 22, 2022.
  • A custom Linux baseline can specify alternative patch repositories without permanently changing the operating system default repositories.
  • When a Linux baseline specifies alternative repositories, only those repositories are used; include default repositories explicitly if they are also needed.
  • Approved patches are installed by an Install operation, while Scan only reports patch compliance.
  • Patch repository: A source location from which an operating-system package manager obtains updates.

Patch compliance and remediation

Patch Manager complianceThe reported state of a node compared with required patching or configuration criteria. compares a node with its approved patch baselineRules and explicit lists that determine approved and rejected patches.. Scanning measures that state; installation changes the node toward the approved patch set.

How it works

  1. Run AWS-RunPatchBaseline with Operation: Scan to identify approved applicable missing updates. Scan does not install or reboot.
  2. Run Operation: Install to apply approved applicable missing updates. Review command output when a patch reports FAILED.
  3. Choose reboot behavior. The default RebootIfNeeded reboots after installing patches or detecting pending-reboot patches during Install.
  4. If using NoReboot, arrange a later reboot and Scan. A baseline snapshotThe approved patch set captured for a patching operation. is the approved patch set captured for an operation; use a consistent snapshot ID when patching nodes together outside a maintenance window.
State Meaning
MISSING Approved patch is absent
FAILED Approved patch could not be installed
INSTALLED_PENDING_REBOOT Node has not rebooted since installation

Exam cues

  • If the question says report missing updates without changing nodes, think Scan, because it measures compliance only.
  • If the question says approved patches are missing, think Install, because it attempts to apply them.
  • If the question says installation failed, think command output, because it contains the information needed to diagnose the failure.

Remember

  • Default RebootOption: RebootIfNeeded.
  • NoReboot prevents OS restarts by Patch Manager; services can still restart.
  • INSTALLED_PENDING_REBOOT is noncompliant.
  • A maintenance window supplies the patch snapshot ID automatically.

Trap: A compliant scan means the node meets the selected baseline. Scan never installs patches.

Trap: RebootIfNeeded does not evaluate each patch vendor’s reboot requirement.

Da documentação da AWS (13)
  • AWS-RunPatchBaseline performs patch operations using the applicable patch baseline.
  • The Scan operation reports approved applicable missing patches without installing updates or rebooting nodes.
  • The Install operation attempts to install approved applicable missing patches and reports patch compliance.
  • RebootOption defaults to RebootIfNeeded.
  • RebootIfNeeded reboots after installing patches or finding INSTALLED_PENDING_REBOOT patches during Install; it does not test whether an individual patch requires reboot.
  • NoReboot prevents operating system restarts by Patch Manager.
  • NoReboot does not prevent service restarts during patching.
  • MISSING means an approved patch is not installed.
  • FAILED means an approved patch could not be installed; AWS directs operators to review command output.
  • INSTALLED_PENDING_REBOOT means the node has not rebooted since installation and is noncompliant.
  • After NoReboot installation, rebooting and running Scan updates a successfully patched node to Compliant.
  • Inside a maintenance window, Patch Manager supplies a snapshot ID; outside one, AWS recommends a shared custom snapshot ID for nodes patched together to use a consistent approved patch set.
  • Baseline snapshot: The approved patch set captured for a patching operation.

Maintenance windows

A maintenance window defines when Systems Manager should schedule maintenance that could disrupt a workload. It separates the allowed execution period from the document that performs the work.

How it works

  1. Set the schedule, duration and cutoffThe time before a maintenance window ends when scheduling of new tasks stops.. Cutoff is the number of hours before the end when new tasks stop being scheduled.
  2. Register targets and tasks. A patching task can run AWS-RunPatchBaseline against managed nodes.
  3. Assign task priorities. Tasks sharing a priority run in parallel; the next priority waits for earlier tasks to reach a final state.
  4. Choose cutoff behavior. CONTINUE_TASK is the default. CANCEL_TASK has different behavior for Run Command and the other task types.
Task type Executes
Run Command A command document on managed nodes
Automation An Automation runbookAn ordered set of maintenance or deployment actions executed by Systems Manager Automation.
Lambda A function
Step Functions A Standard workflow, not Express

Exam cues

  • If the question says patch only in an approved maintenance period, think a maintenance window, because its schedule controls task start times.
  • If the question says run tasks in an ordered sequence, think task priorities, because later priorities wait for earlier tasks to finish.
  • If the question says one patch configuration must span accounts and Regions, think a Quick Setup patch policy, because a maintenance window has one account-Region scope.

Remember

  • Cutoff stops new task scheduling; running tasks continue by default.
  • Same priority means parallel execution.
  • A final state includes failure, so priority alone does not make success conditional.
  • For Run Command, CANCEL_TASK attempts cancellation but cannot guarantee the underlying process stops.

Trap: A window end time is not a guarantee that every process has stopped. Account for duration and cutoff behavior.

Da documentação da AWS (8)
  • Maintenance Windows schedules potentially disruptive maintenance on selected resources.
  • A maintenance window defines a schedule, duration, targets and registered tasks.
  • Cutoff specifies how many hours before the window ends Systems Manager stops scheduling new tasks.
  • At cutoff, CANCEL_TASK stops new invocations for Automation, Lambda and Step Functions; for Run Command it attempts cancellation without guaranteeing the underlying process stops.
  • AWS-RunPatchBaseline can run as a maintenance-window task to patch managed nodes.
  • A maintenance window targets nodes in one account-Region pair; Quick Setup patch policies support multiple account-Region pairs.
  • A patch baseline defines the approval rules used to select patches for an operation.
  • The AWS-owned Automation runbook AWSEC2-PatchLoadBalancerInstance removes an instance from its Classic Load Balancer, ALB or NLB target group, waits for connection draining, installs OS updates, and registers the instance again.

Session Manager

Session Manager opens remote sessions to managed nodes through the agent connection. Native sessions remove the need for an inbound SSH port, SSH keys or a bastionA separate host used to reach machines in a private network., a separate host used to reach private machines.

How it works

  1. Use IAM policies to authorize the user and target node. IAM is the permissions system that controls AWS actions.
  2. Open a browser or CLI session. The channel uses TLSA protocol that encrypts data exchanged over a network connection. 1.2, encryption for data in transit, with optional additional KMS encryption.
  3. Configure S3 or CloudWatch Logs for native session data. CloudTrail captures session API calls; EventBridge can respond to those API events.

Exam cues

  • If the question says shell access without inbound ports, think native Session Manager, because the session uses the managed-node channel.

Remember

  • Linux and macOS default to ssm-user, which is added to sudoers, the configuration granting privileged command access.
  • CloudTrail API records and session content logs are different outputs.
  • A bastion host (jump box) is a single, internet-exposed access point to a private network; Session Manager removes the need for it.

Trap: SSH and port-forwarding session content cannot be logged by Session Manager. Use native sessions when a command transcript is required.

Da documentação da AWS (15)
  • Session Manager provides browser or CLI access to managed nodes without opening inbound ports, maintaining bastions or managing SSH keys for its native sessions.
  • IAM policies control which users may start sessions and which managed nodes they can access.
  • Session Manager encrypts the client-to-node channel with TLS 1.2 and can add KMS encryption.
  • Native session data can be logged to S3 or CloudWatch Logs, with optional KMS encryption.
  • CloudTrail records Session Manager API calls.
  • Session content logging is unavailable for Session Manager port-forwarding and SSH sessions.
  • EventBridge can use CloudTrail StartSession, ResumeSession and TerminateSession API events; an exit command inside a session does not itself produce a detected API event.
  • A maintenance window can schedule AWS-RunPatchBaseline for managed nodes outside business hours.
  • Bastion: A separate host used to reach machines in a private network.
  • TLS: A protocol that encrypts data exchanged over a network connection.
  • sudoers: The operating-system configuration that grants users permission to run privileged commands.
  • A bastion host, or jump box, is a server that is the single access point from an external network to resources in a private network, and its exposure to the internet is a security risk.
  • Default Host Management Configuration supplies node permissions without individual instance profiles in each account and Region where enabled.
  • DHMC requires EC2 instances to use IMDSv2.
  • DHMC requires SSM Agent 3.2.582.0 or later.

Parameter Store tiers and policies

Parameter Store keeps applicationA logical group of AWS resources operated together as one workload. configuration as named values. The tier determines storage limits and policy support; the parameter type determines how the value is represented.

How it works

  1. Choose String for text, StringList for a comma-separated list, or SecureString for an encrypted value.
  2. Choose the standard or advanced tier. Upgrade a standard parameter if its size, policy or sharing needs require advanced.
  3. Update the value to create a numbered version. Request a specific version when a deployment must retain a reviewed value.
  4. Attach parameter policies to an advanced parameter when needed.
Property Standard Advanced
Value size 4 KB 8 KB
Parameters per account and Region 10000 100000
Storage charge No additional charge Charged
Parameter policies Unsupported Supported
Cross-account AWS RAM sharing Unsupported Supported

Exam cues

  • If the question says configuration needs parameter policies, think the advanced tier, because standard parameters do not support policies.
  • If the question says credentials require automatic rotation, think Secrets Manager, because AWS recommends it for that requirement.
  • If the question says reuse configuration across accounts, think advanced parameters with AWS RAM, because standard parameters cannot use that sharing feature.
  • If the question says an AMIA template used to launch an EC2 instance with its operating system and software. parameter is needed in multiple Regions, think a parameter in each Region, because parameters belong to one account and Region.
  • If the question says heavy parameter reads are throttled, think higher throughput, because it raises the API transaction quota at extra charge.

Remember

  • Version one is initial; edits increment it.
  • No version specified means latest; a nonexistent version causes an error.
  • Private subnets reach Systems Manager through interface VPC endpointsA private network entry point for connecting to a supported service through AWS PrivateLink..
  • Hierarchies: up to 15 levels; GetParametersByPath reads everything under a level.

Trap: Shrinking an advanced value does not permit downgrading the parameter.

Da documentação da AWS (20)
  • Parameter Store supports String, comma-separated StringList and encrypted SecureString values.
  • Standard parameter values can be up to 4 KB.
  • The standard tier supports 10000 parameters per account and Region.
  • Standard parameters have no additional storage charge.
  • Standard parameters do not support parameter policies.
  • Advanced parameter values can be up to 8 KB.
  • The advanced tier supports 100000 parameters per account and Region.
  • Advanced parameters incur charges.
  • Advanced parameters support parameter policies.
  • A standard parameter can become advanced; an advanced parameter cannot be changed back to standard.
  • Cross-account sharing through AWS RAM requires advanced parameters.
  • Parameter values begin at version one.
  • Each parameter value edit increments the version.
  • Requests without a parameter version use the latest version.
  • Requesting a nonexistent parameter version fails rather than falling back to latest.
  • AWS recommends Secrets Manager for credentials requiring automatic rotation.
  • Parameter Store parameters belong to one account and Region, so each Region needs its own parameter, for example with that Region's AMI ID.
  • Parameter Store has a default throughput quota for API transactions per second; higher throughput can be turned on for heavier workloads, at extra charge.
  • Systems Manager can be reached through interface VPC endpoints (AWS PrivateLink), so managed nodes and applications in a VPC need no public endpoint access.
  • Parameter Store names can form hierarchies of up to 15 levels separated by forward slashes, and GetParametersByPath returns all parameters under a chosen level.

SecureString encryption

A SecureStringA Parameter Store value encrypted with an AWS KMS key. is a Parameter Store value encrypted with a symmetric AWS KMS key, a key that uses the same secret material to encrypt and decrypt. The parameter name and description remain readable.

How it works

  1. Choose the default AWS managed aws/ssm key or a customer managed keyA KMS key whose permissions and lifecycle are controlled by the customer., whose policy you control.
  2. For standard values, Parameter Store calls KMS Encrypt directly. For advanced values, it uses envelope encryptionEncryption where a data key protects the value and a KMS key protects that data key.: a data key encrypts the value and a KMS key protects that data key.
  3. Retrieve with GetParameter and WithDecryption: true when plaintext is needed. Authorize both parameter retrieval and kms:Decrypt.
  4. For a Secrets Manager reference, use /aws/reference/secretsmanager/ followed by the secret name. Parameter Store passes through the request and honors the secret access policies.

Exam cues

  • If the question says parameter retrieval works but decryption fails, think KMS permissions, because retrieval and decryption are separate permissions.
  • If the question says custom key policyThe policy attached to a KMS key that defines who may use it. or cross-account SecureString sharing is required, think a customer managed key, because the aws/ssm key policy cannot be changed.
  • If the question says reuse a parameter-reading integration for a Secrets Manager secret, think the reserved reference path, because Parameter Store supports that pass-through retrieval.

Remember

  • Only symmetric KMS encryption keys are supported.
  • Standard writes need kms:Encrypt; advanced writes need kms:GenerateDataKey; decrypted reads need kms:Decrypt.
  • Cross-account SecureString sharing requires an advanced parameter and separate customer managed key access.
  • Secret references support GetParameter and GetParameters, not recursive path retrieval or Parameter Store history.

Trap: Encrypting a SecureString does not hide its parameter name or description.

Trap: The Secrets Manager reference path does not bypass the secret access policy.

A decrypted read also needs permission to use the KMS key
12CallerGetParameterParameter StoreWithDecryption trueKMSDecrypt authorization
  1. The caller requests the SecureString with WithDecryption and parameter retrieval permission.

  2. Parameter Store uses KMS decryption on behalf of the caller; kms:Decrypt must also be allowed.

Da documentação da AWS (18)
  • SecureString encrypts the parameter value.
  • SecureString parameter names and descriptions remain plaintext.
  • Parameter Store supports symmetric KMS encryption keys, not asymmetric keys.
  • SecureString uses the AWS managed aws/ssm key if no KMS key is specified.
  • Standard SecureString values are encrypted directly with KMS Encrypt.
  • Advanced SecureString uses the AWS Encryption SDK and envelope encryption, with a data key encrypting the value and a KMS key encrypting the data key.
  • GetParameter with WithDecryption true returns the decrypted plaintext value.
  • Reading decrypted SecureString values requires parameter retrieval permission and kms:Decrypt.
  • Standard SecureString encryption requires kms:Encrypt.
  • Advanced SecureString encryption requires kms:GenerateDataKey.
  • The aws/ssm key policy cannot be changed; use a customer managed key for custom key access controls.
  • Sharing a SecureString across accounts requires an advanced parameter and a customer managed KMS key shared separately with consumers.
  • Parameter Store can retrieve Secrets Manager secrets through names beginning /aws/reference/secretsmanager/.
  • Secrets Manager references support GetParameter and GetParameters, but not GetParametersByPath.
  • Secrets Manager references do not support Parameter Store version history.
  • Parameter Store Secrets Manager references honor the secret access policies and do not retain the secret data or metadata.
  • Symmetric key: A cryptographic key that uses the same secret material to encrypt and decrypt.
  • Envelope encryption: Encryption where a data key protects the value and a KMS key protects that data key.

OpsCenter

OpsCenter keeps operational issues in one place as OpsItemsAn operational work record linking an issue with affected resources and investigation context., work records that link an issue to affected AWS resources. An operator can investigate the record and run an Automation runbookAn ordered set of maintenance or deployment actions executed by Systems Manager Automation. to help resolve it.

How it works

  1. Create an OpsItem manually or configure CloudWatch alarms and EventBridge rules to create one when an issue occurs.
  2. Inspect the affected resource, alarm or event details, and the Config, CloudTrail and CloudWatch context available in OpsCenter.
  3. Run an appropriate Automation runbook to help resolve the issue.
  4. Use a deduplication stringAn identifier combined with the initiating resource to suppress duplicate open operational issues., a chosen identifier for repeated reports, to prevent matching open or in-progress OpsItems from multiplying.

Exam cues

  • If the question says one operational record with resource context and remediation, think OpsCenter, because an OpsItem supports investigation and runbook execution.
  • If the question says an alarm should open an operational issue, think an OpsItem alarm action, because CloudWatch ALARM transitions can create OpsItems.
  • If the question says repeated events create duplicate work, think deduplication, because OpsCenter combines the deduplication string with the initiating resource.

Remember

  • OpsItems can be created manually or through alarms and events.
  • Matching open or in-progress OpsItems suppress duplicates.
  • A matching resolved OpsItem allows a new record for recurrence.

Trap: A deduplication string does not permanently silence an issue. A new item can be created after the old one is resolved.

Da documentação da AWS (7)
  • OpsCenter centralizes investigation and resolution of operational work items called OpsItems.
  • CloudWatch alarms entering ALARM and EventBridge rules can create OpsItems automatically; users can also create them manually.
  • OpsItems show affected resources and alarm or event context, and OpsCenter brings together Config, CloudTrail and CloudWatch information for investigation.
  • OpsCenter can run Systems Manager Automation runbooks to help resolve an OpsItem.
  • OpsItem deduplication uses a hash of the deduplication string and the resource that initiated the OpsItem.
  • If the matching OpsItem is open or in progress, deduplication prevents creation of another matching item.
  • If the matching OpsItem is resolved, another request can create a new OpsItem.

Incident Manager

Incident Manager coordinates people, collaboration and automated response when an incident occurs. It closed to new customers in November 2025; existing enabled accounts can continue using it.

How it works

  1. For an existing enabled account, define a response planAn incident template specifying responders, automated actions and collaboration tools. with responders, automated actions and the collaboration tool.
  2. Connect CloudWatch alarms or EventBridge events to the response plan so detection can create an incident.
  3. Engage contacts or the on-call scheduleA set of rotations that identifies which responder should be notified at a given time., which identifies who should respond at a given time. An escalation pathA notification sequence that engages additional responders during an incident. moves notification to additional responders when needed.
  4. Use Automation runbooksAn ordered set of maintenance or deployment actions executed by Systems Manager Automation. and chat collaboration during the response. After resolution, complete a post-incident analysis to identify improvements and follow-up actions.

Exam cues

  • If the question says an existing customer needs on-call engagement plus automated response, think an Incident Manager response plan, because it connects responders, collaboration and actions.
  • If the question says the response process needs improvement after resolution, think post-incident analysis, because it records improvements and follow-up actions.

Remember

  • Closed to new customers on November 7, 2025.
  • Existing customers can continue in accounts where Incident Manager was already enabled.
  • Response plans define the response; engagement and escalation determine whom to notify.

Trap: An exam description may reflect availability before November 2025. New adoption after the closure date is unsupported.

Da documentação da AWS (7)
  • Incident Manager stopped accepting new customers on November 7, 2025; existing customers can continue using accounts where it was already enabled.
  • An Incident Manager response plan defines responders, automated response actions and collaboration tools for an incident.
  • CloudWatch alarms and EventBridge events can create incidents using response plans.
  • Engagement plans specify contacts or on-call schedules and an escalation path to notify responders.
  • Incident Manager can initiate Systems Manager Automation runbooks for response actions.
  • Incident Manager integrates with Amazon Q Developer in chat applications for responder collaboration.
  • Post-incident analysis identifies response improvements and recommended follow-up actions after resolution.

Application Manager

Systems Manager ApplicationA logical group of AWS resources operated together as one workload. Manager collects operations information for resources treated as one application. It helps existing users investigate problems with application context and run remediation. In July 2026 it entered maintenance modeA service state that preserves existing access and support without new features or Region expansion.: accounts that had not used it before July 30, 2026 cannot start using it, while existing customers can continue in established accounts. Purpose · Availability change

How it works

  1. For an established account, import supported resource metadataDescriptive information such as installed software versions or file attributes rather than file contents., meaning descriptive information about resources. Sources include CloudFormation stacksA group of AWS resources created and managed together through CloudFormation., Resource GroupsA saved collection of resources selected by a query such as resource types and tags., and ECS or EKS clusters. A stack is a set of resources managed from a CloudFormation template.
  2. Group resources into an application, a logical collection operated as a unit. A tagA key and optional value attached to a resource to identify or categorize it. is a key-value labelA user-defined name attached to a particular parameter version.. Tag-based Resource Groups select by resource type and tag filtersA resource-selection condition matching a tag key and optionally its values. within a Region. Application grouping · Resource selection
  3. Configure the source services so their monitoring, complianceThe reported state of a node compared with required patching or configuration criteria., and operational records are available in the application view.
  4. Investigate the affected resource and run an authorized Automation runbookAn ordered set of maintenance or deployment actions executed by Systems Manager Automation., a sequence of operational actions, when remediation is needed. CloudFormation integration also exposes stack and template management.
Information or action Source Meaning
Alarms and application logs CloudWatch Monitoring and diagnostic data
Resource compliance AWS Config Whether resource configuration meets rules
Desired-state compliance State Manager Whether a saved associationA saved configuration binding a document and its inputs to targets and an execution schedule.’s configuration is satisfied
OpsItemsAn operational work record linking an issue with affected resources and investigation context. OpsCenter Operational work records for investigation
Remediation Automation runbooks Authorized actions against applicable resources

An association is a saved desired configuration linked to targets. Configure those associations and the other source services before expecting their information to appear. CloudWatch Logs integration applies to applications, not clusters. EKS visibility covers hosting infrastructure, not individual Pods or containers.

Exam cues

  • If the question says an existing user needs operational context across resources belonging to one application, think Application Manager, because it brings the source services’ information together.
  • If the question says configure a new account after July 2026, think an alternative matched to the use case, because Application Manager is closed to new accounts.
  • If the question says inspect EKS Pods in Application Manager, think the infrastructure boundary, because Pod and container operations are outside its view.
  • If the question says remediate from the application view, think an authorized Automation runbook, because Application Manager uses Systems Manager execution capabilities.

Remember

  • New-account access closed July 30, 2026. Existing users retain access; no new features or Regions are planned.
  • Existing awsApplication tags and AppManager groups remain usable.
  • Alternatives: Resource Groups for Regional grouping; Resource Explorer for cross-account and cross-Region discovery; CloudWatch Application Signals for application observability.

Trap: Maintenance mode is not a shutdown. Existing applications and the underlying Systems Manager APIs continue to work; deleting application tags is not required.

Da documentação da AWS (16)
  • Application Manager aggregates AWS resource operations information in the context of an application, a logical resource group operated as a unit.
  • From July 30, 2026, accounts that had not previously used Application Manager cannot access it; existing customers can continue in established accounts.
  • Application Manager is in maintenance mode: AWS maintains availability and support but adds no new features or Regions.
  • The Application Manager availability change does not affect APIs because the console capability uses existing Systems Manager APIs.
  • Application Manager can import resource metadata from CloudFormation stacks, Resource Groups, ECS clusters, and EKS clusters.
  • Existing users can create an application and add resources or resource groups to provide application context.
  • Tag-based Resource Groups queries select resources by resource type and tag filters.
  • AWS Resource Groups provides named resource groupings based on tags in a single Region.
  • Configure CloudWatch logs and alarms to display their information in Application Manager; the CloudWatch Logs integration applies to applications, not clusters.
  • AWS Config supplies resource-rule compliance, while State Manager associations supply desired-state compliance for Application Manager.
  • OpsCenter supplies application operational work records called OpsItems.
  • Application Manager can execute predefined Automation runbooks against applicable application resources; runbook permissions must be configured.
  • Application Manager integrates with CloudFormation to create and manage stacks and templates; stack changes are reflected in both consoles.
  • Application Manager shows EKS hosting infrastructure but does not manage or show operations information for EKS Pods or containers.
  • Existing awsApplication tags and AppManager resource groups remain usable and do not need removal or replacement because of maintenance mode.
  • AWS recommends Resource Groups for Regional organization, Resource Explorer for discovery across accounts and Regions, and CloudWatch Application Signals for application observability.

Cross-account and Region automation

Multi-account and Region Automation runs a centrally selected runbookAn ordered set of maintenance or deployment actions executed by Systems Manager Automation. in other AWS locationsOne AWS account in one AWS Region used as an Automation execution location.. A location is an account-Region pair; each location can contain multiple target resources.

How it works

  1. Create AWS-SystemsManager-AutomationAdministrationRole in the central account. The caller needs iam:PassRole, permission to pass that role to Automation.
  2. Create AWS-SystemsManager-AutomationExecutionRole in every target account. CloudFormation StackSets can deploy the role across those accounts.
  3. Select accounts or organizational unitsA group of AWS accounts within an AWS Organizations hierarchy. (OUs), Regions and resource targets. An OU is a group of accounts in AWS Organizations. Share a custom runbook with every target account.
  4. Set location concurrencyThe number of target invocations allowed to run at the same time. across account-Region pairs and resource concurrency and error controls within each pair.
  5. Monitor the parent execution for child execution IDs. Read each child execution in its own account and Region for detailed output.

Exam cues

  • If the question says centrally remediate resources across accounts and Regions, think multi-account and Region Automation, because the central role can reach target execution roles.
  • If the question says only a few locations may change at once, think location concurrency, because it limits account-Region pairs.
  • If the question says only a few instances may change in each location, think resource concurrency, because it limits targets within that pair.

Remember

  • Administration role: central account. Execution role: every target account.
  • OU targeting requires the management account or a delegated administrator.
  • Custom runbooks must be shared with all targets.

Trap: Location concurrency does not limit how many resources run inside a location. Configure resource rateA schedule expression specifying a repeating time interval. controls too.

Automation assumes execution roles in target accounts
Central accountTarget account A, Region ATarget account B, Region B12233Automationrunbook and locationsAdmin rolecentral accountExecution roletarget accountTarget nodeschild executionExecution roletarget accountTarget nodeschild execution
  1. Start the central runbook using the administration role.

  2. Use the administration role to assume the execution role in each target account.

  3. Run child automations on the selected resources in each target account and Region.

Da documentação da AWS (11)
  • Systems Manager Automation can run from a central account across target AWS accounts, organizational units and Regions.
  • AWS-SystemsManager-AutomationAdministrationRole belongs in the central account and permits assuming target execution roles.
  • Every target account requires AWS-SystemsManager-AutomationExecutionRole; AWS recommends CloudFormation StackSets to deploy it.
  • The caller needs iam:PassRole for the central Automation administration role.
  • Targeting organizational units requires a management or delegated administrator account that can list accounts in the OUs.
  • A custom runbook must be shared with every target account.
  • Location concurrency limits simultaneous account-Region pairs.
  • Resource rate controls limit concurrent targets and tolerated errors inside each account-Region pair.
  • The parent execution exposes child Automation execution IDs; child output must be retrieved in its target account and Region.
  • Organizational unit: A group of AWS accounts within an AWS Organizations hierarchy.
  • iam:PassRole: Permission for a caller to pass an IAM role to an AWS service.

Quick Setup patch policies

A Quick Setup patch policy is a central patch configuration for chosen accounts and Regions. It combines baselineRules and explicit lists that determine approved and rejected patches. selection with scan and install schedules so the organization can manage them together.

How it works

  1. Choose the target accounts or organizational unitsA group of AWS accounts within an AWS Organizations hierarchy. and Regions, or use a single local account.
  2. Choose a predefined or custom patch baseline for each operating system. Patch-policy operations do not use patch groupsNodes whose patch-group tag value is registered with an operating-system-specific patch baseline..
  3. Set scanning and installation schedules separately. The policy runs AWS-RunPatchBaseline.
  4. Inspect complianceThe reported state of a node compared with required patching or configuration criteria. in each account-Region pairOne AWS account in one AWS Region used as an Automation execution location.. Changes to custom baselines synchronize with Quick Setup once an hour.

Exam cues

  • If the question says central patch setup across an organization, think a Quick Setup patch policy, because it supports accounts, OUs and Regions in one configuration.
  • If the question says scan frequently but install less often, think separate policy schedules, because scan and install timing can differ.
  • If the question says a patch-group tagA key and optional value attached to a resource to identify or categorize it. does not select the policy baseline, think the policy baseline settings, because policy operations do not use patch groups.

Remember

  • Select baselines per operating system.
  • Compliance remains per account-Region pair.
  • Custom baseline synchronization: once an hour.
  • On-demand Patch now targets one account-Region pair at a time.

Trap: An organization-wide configuration does not make every patching feature organization-wide. Patch now still has one account-Region scope.

Da documentação da AWS (8)
  • Quick Setup patch policies configure patching across selected accounts, organizational units and Regions, or one local account.
  • Patch policies can configure separate schedules for scanning and installation.
  • A patch policy selects an AWS predefined or custom baseline for each operating system.
  • Patch policy operations do not use patch groups.
  • Patch policy operations run the AWS-RunPatchBaseline command document.
  • Patch policy compliance is reported separately for each account-Region pair.
  • Patch now remains limited to one account-Region pair at a time.
  • Changes to custom patch baselines used by Quick Setup are synchronized once an hour.

Inventory and compliance reporting

Inventory records what is installed and configured on managed nodes. ComplianceThe reported state of a node compared with required patching or configuration criteria. records whether patching and associationsA saved configuration binding a document and its inputs to targets and an execution schedule. meet their expected state; compliance history requires AWS Config.

How it works

  1. Create an Inventory association using AWS-GatherSoftwareInventory and choose the metadataDescriptive information such as installed software versions or file attributes rather than file contents., descriptive details such as software versions, to collect. Add custom JSON metadata when built-in fields are insufficient.
  2. Set the collection interval. Inventory gathers metadata, such as file attributes and software versions, rather than file contents.
  3. Create resource data syncA configuration that copies collected inventory data to a central S3 bucket. configurations to copy collected data into a central S3 bucket. Use Athena to query data from accounts and Regions.
  4. Use Compliance for current patch and association status. Enable AWS Config recording for SSM:PatchCompliance and SSM:AssociationCompliance to retain changes over time.
  5. Use the Patch Manager integration to send noncompliant-node findingsA recorded security check result or security-related detection. to Security Hub CSPM. A finding is the recorded result of a security check.

Exam cues

  • If the question says which software versions are installed across the fleet, think Inventory, because it gathers applicationA logical group of AWS resources operated together as one workload. metadata.
  • If the question says query inventory across accounts and Regions, think resource data sync plus Athena, because S3 holds the aggregated data.
  • If the question says show how compliance changed over time, think AWS Config, because Compliance itself shows current state.

Remember

  • Minimum collection interval: 30 minutes.
  • Association compliance includes status and severity.
  • Patch Manager sends one finding per noncompliant managed nodeA machine configured to receive Systems Manager management actions through SSM Agent. to Security Hub CSPM.
  • A subsequent compliant node result does not automatically update that finding.

Trap: Collecting file inventory does not upload the files themselves.

Trap: A successful remediation and a resolved Security Hub finding are separate states; resolve the finding after checking compliance.

Centralize collected inventory before querying across locations
12Inventorycollected node metadataS3 bucketresource data syncAthenaquery collected metadata
  1. Resource data sync copies collected inventory from source accounts and Regions to the central S3 bucket.

  2. Athena queries the aggregated inventory data in S3.

Da documentação da AWS (13)
  • Inventory collects metadata such as application versions, file attributes, network configuration, Windows updates, services and tags rather than file contents.
  • Custom inventory is JSON metadata placed on a managed node for collection.
  • The minimum Inventory collection interval is 30 minutes.
  • AWS-GatherSoftwareInventory uses a State Manager association to collect inventory metadata.
  • Inventory resource data sync continuously updates a central S3 bucket as new collected inventory data is discovered.
  • Inventory data from multiple accounts and Regions can be queried with Athena.
  • Systems Manager Compliance reports current patch, association and custom compliance data.
  • Association compliance includes Compliant or Non-compliant status and an assigned severity.
  • Enable SSM:PatchCompliance and SSM:AssociationCompliance recording in AWS Config to retain compliance history and track changes.
  • Patch Manager can send one Security Hub CSPM finding per noncompliant managed node, including patch summary results.
  • Patch Manager does not update a Security Hub CSPM finding after the node becomes compliant; findings can be resolved manually.
  • Metadata: Descriptive information such as installed software versions or file attributes rather than file contents.
  • Finding: A recorded security check result or security-related detection.

Cross-account parameter sharing

Cross-account Parameter Store sharing lets applicationsA logical group of AWS resources operated together as one workload. read configuration owned by another AWS account. AWS Resource Access Manager (RAM) shares the existing advanced parameter with consumers, the accounts allowed to use it. The owner keeps control of updates and deletion.

How it works

  1. In the owning account, create or upgrade to an advanced parameter. Only the owner can share it.
  2. Add the parameter to a RAM resource shareAn AWS RAM configuration listing shared resources, allowed consumers, and their permissions., a record of the resource, allowed consumers, and permissions. Select accounts or enabled AWS Organizations targets. An organizational unitA group of AWS accounts within an AWS Organizations hierarchy. (OU) is a group of accounts. Consumers accept an invitation unless sharing within the organization is enabled.
  3. Choose read-only accessPermission to retrieve a shared parameter without updating, deleting, or resharing it. with or without version history. A recipient can neither update, delete, nor reshare the parameter.
  4. For SecureString, an encrypted parameter value, use a customer managed KMS key and authorize that key separately. The owner’s key policyThe policy attached to a KMS key that defines who may use it. and the consumer’s IAM permissions must permit key use. Plaintext retrieval requires kms:Decrypt.
  5. Retrieve with GetParameter using the full ARNAn AWS resource identifier that includes the owning account, Region where applicable, and resource path., the AWS identifier containing the Region, owner account, and parameter path. Set WithDecryption: true for plaintext. Use describe-parameters --shared for discovery; the consumer’s My parameters page does not list these values. Read API · Cross-account key access
RAM permission setAn AWS RAM collection of allowed actions granted to a resource-share consumer. Allowed reads History
AWSRAMDefaultPermissionSSMParameterReadOnly DescribeParameters, GetParameter, GetParameters No
AWSRAMPermissionSSMParameterReadOnlyWithHistory The same actions plus GetParameterHistory Yes
Integration Shared-parameter support
CloudFormation template parameter inputs Supported
CloudFormation dynamic references, values resolved from an external store in resource properties Unsupported
EC2 launch templates and RunInstances image references Supported
Parameters and Secrets Lambda extension; Automation runbookAn ordered set of maintenance or deployment actions executed by Systems Manager Automation. retrieval Supported
Run Command parameter references Unsupported
CodeBuild or App Runner environment values; ECS secret values Unsupported

Exam cues

  • If the question says centrally maintained configuration must be read across accounts, think an advanced parameter shared through RAM, because sharing preserves ownership and grants consumer reads.
  • If the question says a shared SecureStringA Parameter Store value encrypted with an AWS KMS key. cannot be decrypted, think separate KMS authorization, because sharing the parameter does not share the encryption key.
  • If the question says read access works but history is denied, think the history-enabled RAM permission set, because default read-only access omits GetParameterHistory.
  • If the question says use a shared parameter in CloudFormation, think template parameter inputs, because dynamic references do not support shared parameters.

Remember

  • Standard parameters cannot be shared through this feature.
  • AWS managed KMS keys cannot be shared; encrypted sharing needs a customer managed keyA KMS key whose permissions and lifecycle are controlled by the customer..
  • Use the full parameter ARN for a cross-account read.
  • The owner pays storage; each consumer pays its API use and has separate retrieval throughput.
  • Removing the parameter from the share ends consumer access.

Trap: A RAM share grants parameter access, not decryption by itself. Check the KMS key policy, consumer IAM permissions, and decryption request separately.

Parameter reads and decryption require separate cross-account authorization
Owner accountConsumer accountsharereadkey policy and consumer IAMParameter Storeadvanced parameterRAM shareread-only accessApplicationGetParameter: full ARNKMS keyDecrypt permission
Da documentação da AWS (19)
  • Only the account that owns a parameter can share it; a recipient cannot reshare it.
  • Cross-account Parameter Store sharing through AWS RAM requires advanced parameters.
  • An AWS RAM resource share specifies the parameters, consumer accounts or organization targets, and read-only permission set.
  • Consumers accept a resource-share invitation unless sharing within their AWS organization is enabled; organization or OU sharing requires that organization-sharing setting.
  • Recipients can read shared parameters but cannot update or delete them.
  • A shared SecureString must use a customer managed KMS key with separately authorized cross-account key access; AWS managed keys cannot be shared.
  • Cross-account KMS key use needs permission in the owning key policy and in an IAM policy for the external identity.
  • Retrieving SecureString plaintext requires kms:Decrypt permission on its KMS key.
  • GetParameter requests for a parameter shared from another account must identify it with its full ARN.
  • GetParameter WithDecryption returns decrypted SecureString values and is ignored for String and StringList.
  • describe-parameters --shared lists parameters shared with the consumer; they do not appear on the consumer My parameters page.
  • AWSRAMDefaultPermissionSSMParameterReadOnly permits DescribeParameters, GetParameter, and GetParameters without GetParameterHistory.
  • AWSRAMPermissionSSMParameterReadOnlyWithHistory adds GetParameterHistory to the read-only permission set.
  • Shared parameters are supported as CloudFormation template parameter inputs, but not CloudFormation dynamic references.
  • Shared parameters support EC2 launch templates, EC2 RunInstances ImageID references, the Parameters and Secrets Lambda extension, and retrieval in Automation runbooks.
  • Shared parameters are unsupported as Run Command command parameters, CodeBuild or App Runner environment variable values, and ECS secret values.
  • Removing a parameter from a resource share ends the consumer account access to that parameter.
  • Shared-parameter retrieval throughput is enforced separately for each consuming account.
  • The owner pays for advanced parameter storage, and each consumer pays for its API use of shared advanced parameters.

Parameter Store operational controls

Systems Manager Parameter Store uses hierarchies, policies and version history to organize parameters and manage their lifecycle. These controls affect how applicationsA logical group of AWS resources operated together as one workload. retrieve values and how updates behave over time.

How it works

  1. Organize parameter names in a slash-separated hierarchyA slash-separated parameter name structure that organizes related configuration values.. Use GetParametersByPath with recursive retrievalReading parameter values from the requested path and its nested paths. to read parameters below a path.
  2. Use policies on an advanced parameter to delete it after an expiration timestamp or emit EventBridge events. ExpirationNotification alerts before or after expiration; NoChangeNotification alerts when the parameter has remained unchanged for the specified period. Expiration is processed asynchronously through periodic scans.
  3. When updating policies, supply the complete desired policy array: a new array replaces the existing policies.
  4. Parameter Store retains version history. Once the history is full, the oldest version is normally removed to make room for a new one. If the oldest version has a labelA user-defined name attached to a particular parameter version., the update fails; move the label to a newer version before retrying.

Exam cues

  • If the question says an application needs parameters below an environment path, think a hierarchy and recursive GetParametersByPath, because it retrieves parameters below the requested path.
  • If the question says to alert on a parameter that has not changed, think NoChangeNotification, because it emits an EventBridge event after the specified unchanged period.
  • If the question says an update fails with a full version history, think a label on the oldest version, because that label prevents its removal.

Remember

  • Hierarchies support up to 15 levels.
  • An advanced parameter supports up to ten policies.
  • Parameter Store retains up to 100 versions.
  • A standard parameter can become advanced; an advanced parameter cannot be changed back to standard.
  • Session Manager idle timeout: 20 minutes by default, one to 60 minutes; resizing or reconnecting can reset it. Maximum session duration is a separate limit.
  • Restricted command access: custom Session documents with sessionType: InteractiveCommands, started through the CLI.
  • Run AsA session setting that selects an existing Linux or macOS operating-system user. selects an existing Linux or macOS user (the SSMSessionRunAs tagA key and optional value attached to a resource to identify or categorize it. wins over the account preference) and fails rather than falling back to ssm-user.

Trap: Expiration deletes the parameter; ExpirationNotification emits an event. Also, supplying a new policy array replaces the existing array rather than adding to it.

Da documentação da AWS (26)
  • A standard parameter can become advanced; an advanced parameter cannot be changed back to standard.
  • Expiration deletes a parameter after the specified timestamp through asynchronous periodic policy processing.
  • ExpirationNotification emits an EventBridge event before or after the expiration time.
  • NoChangeNotification emits an EventBridge event when a parameter has remained unchanged for the specified period.
  • An advanced parameter can have up to ten policies.
  • Supplying a new policy array replaces the policies already attached to the parameter.
  • Parameter hierarchies use slash-separated names.
  • Parameter hierarchies support up to 15 levels.
  • GetParametersByPath supports recursive retrieval of parameters below a path.
  • Parameter Store retains up to 100 versions, normally deleting the oldest to make room for the next.
  • If the oldest of 100 retained versions has a label, creation of a new version fails instead of deleting the labeled version.
  • A custom Session document using schemaVersion 1.0 and sessionType InteractiveCommands can restrict commands and parameters.
  • InteractiveCommands sessions are supported only through the CLI.
  • IAM policies control which users may start sessions and which managed nodes they can access.
  • If Run As cannot use the selected operating-system user, the session fails without falling back to ssm-user.
  • Session Manager idle timeout defaults to 20 minutes.
  • Session Manager idle timeout can be set from one to 60 minutes.
  • Keyboard input, window resizing and reconnection can reset the Session Manager idle timeout.
  • Maximum session duration is a separate control from idle timeout.
  • On Linux and macOS, Session Manager uses ssm-user by default and adds it to sudoers.
  • Run As chooses an existing Linux or macOS operating-system user using SSMSessionRunAs on the IAM entity before the user configured in session preferences.
  • Tasks run in priority order; tasks with the same priority run in parallel.
  • A later-priority task starts after earlier-priority tasks reach a final state, regardless of whether those tasks succeeded.
  • Cutoff specifies how many hours before the window ends Systems Manager stops scheduling new tasks.
  • CutoffBehavior defaults to CONTINUE_TASK, allowing running tasks to continue after cutoff.
  • Maintenance windows support Run Command, Automation, Lambda and Step Functions Standard tasks; Express workflows are unsupported.

Flashcards

toque para virar
Resposta

·

Todos os 119 flashcards em lista

Operations must patch hundreds of EC2 instances and on-premises servers from one place. Which AWS service fits?

AWS Systems Manager, which manages nodes in AWS and on premises through SSM Agent.

How can admins open a shell on EC2 instances without opening SSH ports or running a bastion host?

Systems Manager Session Manager.

Which AWS service helps a company centrally manage servers across AWS, on premises and other clouds?

AWS Systems Manager centrally manages and operates nodes across these environments.

A company wants one place to view and operate both its EC2 instances and its on-premises servers. Which AWS service fits?

AWS Systems Manager, which manages them all as managed nodes from a centralized view.

What makes an EC2 instance a Systems Manager managed node?

SSM Agent, node IAM permissions and connectivity to Systems Manager. Permissions can come from an instance profile or Default Host Management Configuration.

A private EC2 instance does not appear in Systems Manager. What should the operator check?

Check SSM Agent, node permissions and outbound connectivity. For private endpoints, check ssm, ssmmessages and endpoint security groups allowing port 443 from the node. Do not require ec2messages in Regions launched in 2024 or later.

How do on-premises servers become Systems Manager managed nodes through hybrid activation?

Create an activation ID and code, then register SSM Agent. The activation assigns a service role trusting Systems Manager; registered node IDs start with mi-. An EC2 instance-profile role instead trusts ec2.amazonaws.com.

A Systems Manager EC2 node still uses its instance role after Default Host Management Configuration is enabled. Why?

An instance profile allowing ssm:UpdateInstanceInformation takes precedence. Remove that permission from the profile if the node should use Default Host Management Configuration.

A team wants to manage Greengrass core devices with Systems Manager. Is a generic hybrid activation sufficient?

Follow the Greengrass setup: configure the registration service role and token exchange role, then deploy the SSM Agent component. Greengrass registers the devices automatically.

What are the default expiration and registration limit for a Systems Manager hybrid activation?

24 hours and one registration. Expiration can be up to 30 days ahead, with a registration limit up to 1000 per activation.

An operator is preparing IAM roles for EC2 and hybrid Systems Manager nodes. Which service does each role trust?

The role in an EC2 instance profile trusts ec2.amazonaws.com. The hybrid activation service role trusts ssm.amazonaws.com.

A company must find EC2 instances that Systems Manager is not managing. What fits?

The AWS Config rule ec2-instance-managed-by-systems-manager (or comparing running instances with DescribeInstanceInformation).

A team needs Systems Manager to run bootstrap scripts stored in GitHub or S3. Which document fits?

AWS-RunRemoteScript. It downloads and runs remote scripts on managed nodes. For S3, grant the node profile s3:ListBucket and s3:GetObject.

A team needs Systems Manager to run a command on nodes tagged Env=Prod. How should it target them?

Use Run Command with a command document and a tag target. Targets can also be node IDs or a resource group; tag targeting permits up to five keys and five values per key.

A team needs reusable Systems Manager configuration actions with a fixed reviewed revision. What should it use?

A parameterized JSON or YAML SSM document with an explicit document version. Updating content creates another version; the default version can be changed separately.

A custom Systems Manager command document must run another SSM document from GitHub. Which plugins belong in sequence?

Use aws:downloadContent with sourceType GitHub and repository sourceInfo, then aws:runDocument with documentType LocalPath and the downloaded document path.

A Systems Manager command produces more output than its default response shows. How can a team retain all output and receive failure alerts?

Send complete output to S3 or CloudWatch Logs and configure SNS status notifications. Notifications can describe the command or each node invocation.

A team wants a Systems Manager script to configure nodes after Auto Scaling launch events. Which event integration fits?

Use an EventBridge rule with Run Command as its target to run the command document on managed nodes.

A team wants Systems Manager targets selected by resource type and environment tags in a particular Region. What should it use?

Use a tag-based resource group as the Run Command target. Its query filters resource types and tag values; resource groups are maintained separately by Region.

A CloudWatch alarm notifies an SNS topic when an application on an EC2 instance runs out of memory, and the application must be restarted automatically. What can the topic trigger to do it?

A Lambda function subscribed to the topic that calls SendCommand to run a restart command on the instance with Run Command.

What are the Systems Manager Run Command defaults for max-concurrency and max-errors?

50 concurrent invocations and zero allowed errors. Both accept a number or percentage; zero errors stops further dispatch after the first error.

A Systems Manager Run Command rollout exceeds its error threshold. Are running invocations cancelled?

No. Running invocations complete and may also fail. Set max-concurrency to one when the rollout must proceed serially.

A Systems Manager Run Command rollout sets max-errors to three. At which error does it stop dispatching to new nodes?

The fourth error. max-errors is the number allowed before new dispatch stops.

A team must stop a Systems Manager Run Command rollout after the first failure without other nodes failing concurrently. Which controls fit?

Use max-errors zero and max-concurrency one. The first error stops new dispatch, and serial execution avoids other in-flight invocations.

What does the Systems Manager AWS-UpdateLinuxAmi Automation runbook do?

It uses a source AMI and workspace instance to create a new AMI with Linux package and Amazon software updates.

A team needs Systems Manager to coordinate AWS resource changes and pause for a manager before continuing. Which document type fits?

An Automation runbook with an aws:approve step. A runbook coordinates maintenance actions and pauses for designated approvers.

How does a Systems Manager Automation runbook wait for human approval?

Use aws:approve with designated approvers. The timeout defaults to seven days and can be up to 30 days. An optional SNS notification topic must have a name starting with Automation.

A Systems Manager Automation runbook should perform EC2 instance-type changes under a service role. Which input and runbook fit?

Set AutomationAssumeRole on AWS-ResizeInstance. The runbook uses that role for actions; without it, the runbook uses the initiating user permissions.

A team uses Systems Manager Automation to patch its golden AMI and publish the resulting AMI ID for consumers. How can it do that?

Build the patched AMI and invoke a Lambda function at the end to update a Parameter Store value with the new AMI ID, as in the AWS golden AMI tutorial.

An EC2 instance is unreachable after operating-system changes. Which Systems Manager Automation can attempt offline repair?

AWSSupport-ExecuteEC2Rescue. It stops and backs up the instance, repairs its root volume through a temporary managed helper, then restores and restarts it. Encrypted root volumes are unsupported.

Can a Systems Manager aws:approve step be placed directly in a multi-account and Region Automation?

No. The aws:approve action does not support that execution mode.

A Windows EC2 instance cannot boot or accept RDP after firewall or network changes. Which Systems Manager workflow can attempt offline recovery?

Use AWSSupport-ExecuteEC2Rescue to run repair through a temporary managed helper. EC2Rescue for Windows can diagnose boot, firewall, RDP and network-interface issues; the Automation does not support encrypted root volumes.

An operations team needs scripts on instances to run in a fixed order, stop on failure, and alert by email if a run fails, without custom code. What fits?

A Systems Manager Automation runbook with aws:runCommand steps, plus an EventBridge rule on Failed executions that notifies SNS.

A team wants Systems Manager to run its existing Chef recipes on Linux nodes. What should it use?

A State Manager association with AWS-ApplyChefRecipes. It supports Chef 11 through 18 and cookbook downloads from Git, GitHub, HTTP or S3.

A team needs Systems Manager to reapply instance settings on a schedule and report compliance. Which capability fits?

State Manager. An association specifies the document, parameters, targets and schedule; Compliance reports its status and severity.

A fleet needs recurring configuration and compliance reporting. Which Systems Manager capability maintains that configuration?

State Manager associations. They bind a desired configuration to targets and a schedule and report association compliance.

New Auto Scaling nodes must receive the same CloudWatch agent configuration through Systems Manager. Which setup fits?

A State Manager command-document association targeting the matching node tags. State Manager attempts to run it when new matching managed nodes first come online and reapplies it on schedule.

A Systems Manager command-document association should run only at its next cron interval, including when new nodes match. Which setting fits?

Set ApplyOnlyAtCronInterval. The association then avoids executions triggered by target changes.

Does Systems Manager State Manager guarantee immediate repair whenever a file is manually changed?

No. An association reapplies configuration on its schedule and documented triggers; it is not a promise of immediate file-change detection.

How does a Systems Manager node select its patch baseline when using patch groups?

Its Patch Group or PatchGroup tag value selects the registered baseline for its operating system. If no matching baseline exists, the current OS default applies. A node has one patch group, with one baseline per OS in that group.

A Windows fleet must install approved updates sooner than the Systems Manager predefined baseline allows. Can the team edit AWS-DefaultPatchBaseline?

No. Copy or create a custom baseline with a shorter approval delay. Predefined Windows baselines wait seven days and cannot be customized.

How do auto-approval delay and cutoff date differ in Systems Manager Patch Manager?

A delay waits a chosen number of days after release or update. A cutoff date approves patches released or updated on or before that fixed date. Rejection wins if a patch is both approved and rejected.

A Linux Systems Manager fleet must patch from a corporate repository and its usual repositories. What should its baseline specify?

Use a custom baseline with alternative patch sources, listing both the corporate and usual repositories. Only specified repositories are used during that operation; OS defaults remain unchanged afterward.

An EC2 fleet allows tags in instance metadata. Which Systems Manager patch-group tag key should it use?

PatchGroup without a space. Patch Group cannot be used when instance metadata tags are enabled.

A Systems Manager patch report shows MISSING and FAILED patches. How should an operator remediate them?

Use Install for approved missing patches. For FAILED patches, inspect command output to diagnose installation errors. Scan reports compliance without installing updates.

Which Systems Manager AWS-RunPatchBaseline operation changes installed patches: Scan or Install?

Install applies approved applicable missing updates. Scan only reports compliance and does not install or reboot.

A Systems Manager patch installation uses NoReboot and leaves nodes noncompliant. What should happen next?

Schedule the reboot, then run Scan. INSTALLED_PENDING_REBOOT records that the node has not rebooted since patch installation.

Does the default RebootIfNeeded option in Systems Manager only reboot when a patch vendor requires it?

No. It reboots after installing patches or detecting pending-reboot patches during Install, without testing whether a particular patch requires it.

A team uses Systems Manager Run Command outside a maintenance window to patch a fleet consistently. What should it supply?

A common custom snapshot ID for the nodes patched together. It keeps the operation on a consistent approved patch set.

A team needs Systems Manager patch installation only during an approved maintenance period. Which feature controls execution time?

A maintenance window with a schedule, duration, cutoff, targets and a task such as AWS-RunPatchBaseline. The baseline controls approved patches.

A team must patch EC2 and registered on-premises nodes outside business hours. How can Systems Manager select patches and execution time?

Use Patch Manager baselines to select approved patches, then schedule AWS-RunPatchBaseline in a maintenance window.

Instances behind an ALB must be patched without receiving traffic while they patch, within a defined window. What fits with the least custom work?

Run the AWSEC2-PatchLoadBalancerInstance Automation runbook from a Systems Manager maintenance window. It deregisters each instance, waits for draining, patches it and registers it again.

Engineers need Systems Manager shell access without inbound SSH and with an audit trail. What should be configured?

Use native Session Manager access with IAM authorization. Send session data to S3 or CloudWatch Logs; CloudTrail records session API calls.

Does Systems Manager Session Manager record typed commands in SSH or port-forwarding sessions?

No. Session content logging is unavailable for those sessions. CloudTrail still records Session Manager API calls.

A hybrid fleet needs access without inbound SSH and scheduled patching outside business hours. Which Systems Manager features fit?

Session Manager for native remote access and Patch Manager through a maintenance window for scheduled patching.

Administrators reach private EC2 instances through a public bastion host with SSH open. Which AWS feature removes the inbound port and the bastion?

AWS Systems Manager Session Manager, with sessions authorized by IAM and logged.

An EC2 fleet needs Systems Manager permissions without individual instance profiles. What must be enabled and installed?

Enable Default Host Management Configuration in each account and Region. Instances must use IMDSv2 and SSM Agent 3.2.582.0 or later.

How do Systems Manager Parameter Store standard and advanced parameters differ?

Standard: 4 KB values, 10000 parameters per account and Region, no additional storage charge and no policies. Advanced: 8 KB, 100000 parameters, charges, policies and AWS RAM sharing. Advanced parameters cannot be downgraded.

A deployment must keep using a reviewed Systems Manager parameter value after later edits. How should it retrieve the value?

Specify the parameter version. Value changes create numbered versions; omitting a version retrieves the latest, and requesting a nonexistent version fails.

A deployment needs to store an encrypted credential in AWS Systems Manager Parameter Store. Which parameter type should it use?

SecureString. Parameter Store supports encrypted SecureString values.

Which Systems Manager Parameter Store types represent text, a comma-separated list and an encrypted value?

String for text, StringList for a comma-separated list, and SecureString for an encrypted value.

A template reads a golden AMI ID from Parameter Store and must deploy in three Regions. What is needed?

A parameter in each Region holding that Region's AMI ID; parameters are Regional.

An application reading parameters heavily gets ThrottlingException from Parameter Store. What setting helps?

Turn on higher throughput for Parameter Store (extra charge), or cache the values in the application.

Applications in private subnets with no NAT gateway must read AWS Systems Manager Parameter Store values. What should the VPC add?

An interface VPC endpoint for Systems Manager (AWS PrivateLink).

An application must load every AWS Systems Manager Parameter Store value for /prod/payments/ at startup. Which naming and API fit?

Hierarchical names such as /prod/payments/db-url, read with GetParametersByPath (recursively if needed).

How is a Systems Manager Parameter Store SecureString encrypted, and what permissions does a reader need?

Use a symmetric KMS key, defaulting to aws/ssm. A reader requests WithDecryption and needs permission to retrieve the parameter and kms:Decrypt. Use a customer managed key for custom key-policy controls.

An application can retrieve a Systems Manager SecureString but gets AccessDenied when requesting plaintext. What should the team check?

Check kms:Decrypt access to the parameter encryption key in addition to parameter retrieval permission.

A team needs to share a Systems Manager SecureString with another account. Can it use the default aws/ssm key?

No. Use an advanced parameter with a customer managed KMS key and grant consumers access to that key separately.

How do standard and advanced Systems Manager SecureString encryption differ?

Standard values use KMS Encrypt directly. Advanced values use envelope encryption: a data key encrypts the value and a KMS key protects that data key.

An application already reads Systems Manager parameters but needs a Secrets Manager secret. How can it retrieve the secret through that interface?

Use GetParameter or GetParameters with /aws/reference/secretsmanager/ followed by the secret name. Secret access policies still apply; GetParametersByPath and Parameter Store history are unsupported for these references.

What is Systems Manager OpsCenter, and how are OpsItems created?

OpsCenter centralizes operational issues as OpsItems. They can be created manually or automatically by CloudWatch alarms and EventBridge rules, then investigated and resolved with resource context and Automation runbooks.

A team needs CloudWatch alarms to create operational work records with affected resources and remediation runbooks. Which Systems Manager tool fits?

OpsCenter. An alarm entering ALARM can create an OpsItem with investigation context, and operators can run Automation runbooks.

Repeated events for the same resource should not open duplicate Systems Manager OpsItems. What should the team configure?

A consistent deduplication string. OpsCenter hashes it with the initiating resource and suppresses creation when the matching OpsItem is open or in progress.

A Systems Manager OpsItem was resolved before the same issue recurred. Does deduplication prevent a new OpsItem forever?

No. A matching resolved OpsItem allows a new OpsItem to be created.

An existing Incident Manager customer wants an alarm to notify on-call responders and start a runbook. What should it configure?

An Incident Manager response plan with responder engagement, collaboration and an Automation runbook, triggered by CloudWatch or EventBridge. New customers have not been accepted since November 7, 2025.

An existing Incident Manager customer needs an unanswered incident notification to reach additional responders. Which configuration fits?

An engagement plan with contacts or an on-call schedule and an escalation path.

How can a CloudWatch alarm start an automated response using Systems Manager Incident Manager?

Configure the alarm to create an incident using a response plan. Incident Manager can initiate Systems Manager Automation runbooks for response actions.

A Systems Manager Incident Manager response must notify an on-call team and escalate if needed. What should define the notification path?

Use an engagement plan specifying contacts or on-call schedules and an escalation path.

What does Systems Manager Application Manager organize for an existing customer?

Operations information for a logical group of AWS resources managed as one application, including monitoring, compliance, and operational issues.

An existing Systems Manager Application Manager user wants to see application alarms and logs. What must be configured?

Configure CloudWatch alarms and logs for display in Application Manager. The logs integration applies to applications, not clusters.

An existing Application Manager customer needs alarms, compliance status, and operational work records together for an application. Which source services supply them?

CloudWatch supplies alarms, AWS Config and State Manager supply compliance, and OpsCenter supplies OpsItems. Configure the source integrations.

An existing Application Manager customer wants to remediate a resource problem from the application view. What execution mechanism and access are needed?

Use a Systems Manager Automation runbook with the required runbook permissions for the applicable resources.

An Application Manager user sees an EKS cluster but wants to inspect its individual Pods there. Is that supported?

No. Application Manager exposes EKS hosting infrastructure information, not Pod or container operations information.

An operations team needs application context when reviewing resource health information in Systems Manager Application Manager. What does the service aggregate?

It aggregates AWS resource operations information in the context of an application, a logical resource group operated as a unit.

An existing Systems Manager Application Manager user wants operational work records alongside application information. Which service supplies them?

OpsCenter supplies operational work records called OpsItems.

A team needs to review application resource compliance in Systems Manager Application Manager. Which services supply the compliance information?

AWS Config supplies resource-rule compliance; State Manager associations supply desired-state compliance.

A team must run a Systems Manager Automation runbook across member accounts and Regions. What must it configure?

Create the administration role in the central account and an execution role in every target account. Select target accounts or OUs and Regions; share custom runbooks with every target. OU targeting requires a management or delegated administrator account.

What do location concurrency and resource concurrency control in Systems Manager multi-account Automation?

Location concurrency limits simultaneous account-Region pairs. Resource concurrency limits targets running within each pair; resource error controls also apply within each pair.

A Systems Manager parent Automation shows child execution IDs but no detailed target output. Where should an operator look?

Retrieve the child execution output using its execution ID in the target account and Region.

A central Systems Manager workflow cannot assume the role needed to launch cross-account Automation. Which caller permission should be checked?

Check iam:PassRole on AWS-SystemsManager-AutomationAdministrationRole, along with the administration and target execution role setup.

A team wants one Systems Manager patch configuration across an AWS organization and several Regions. Which feature fits?

A Quick Setup patch policy. It selects accounts or OUs, Regions, per-OS baselines and patch schedules without patch-group mappings.

An organization wants daily Systems Manager patch scans but installations on a separate schedule. What should it configure?

Separate scan and install schedules in a Quick Setup patch policy.

A team creates an organization-wide Systems Manager patch policy. Does that automatically create one combined compliance report for every account and Region?

No. Compliance is reported separately per account-Region pair.

A team edits a custom baseline used by a Systems Manager Quick Setup patch policy. Is the policy baseline updated immediately?

No. Custom baseline updates synchronize with Quick Setup once an hour.

A State Manager association is noncompliant across several managed nodes. Where can the team inspect status and severity?

Systems Manager Compliance. It reports each association compliance status and assigned severity.

What does Systems Manager Inventory collect, and how can it be queried across accounts and Regions?

Node metadata, not file contents. Use resource data sync to centralize collected inventory in S3 and Athena to query it across accounts and Regions. The shortest collection interval is 30 minutes.

An auditor needs historical Systems Manager patch and association compliance changes. Which service must record them?

AWS Config, with SSM:PatchCompliance and SSM:AssociationCompliance enabled. Systems Manager Compliance shows current data.

A Systems Manager node was patched and is compliant, but its Security Hub finding remains open. Why?

Patch Manager does not send a compliant-state update for that finding. Resolve the finding manually after remediation.

A team needs Systems Manager Inventory to include custom equipment attributes absent from built-in metadata. What should it provide?

A custom inventory JSON file on each managed node for Inventory to collect.

A company wants one Parameter Store configuration value read by several AWS accounts without copying it. What sharing setup fits?

Store an advanced parameter in the owning account and add it to an AWS RAM resource share for the consumers. Recipients receive read-only access.

A shared Parameter Store SecureString is readable as encrypted data, but a consumer cannot retrieve plaintext. Which authorization layer needs checking?

Check cross-account KMS access, including the owner key policy, consumer IAM permission for kms:Decrypt, and a request using WithDecryption. The RAM parameter share does not share the key.

Can a Parameter Store SecureString encrypted with the default AWS managed key be shared across accounts through AWS RAM?

Not with that key. Use a customer managed key and separately authorize cross-account key use; the parameter must be advanced.

An application in another AWS account cannot retrieve a shared Parameter Store value using its short path. What identifier must GetParameter use?

The full parameter ARN, including owner account and Region, rather than only its name.

An AWS RAM consumer can read a shared Parameter Store value but cannot read its version history. Which permission set enables history?

AWSRAMPermissionSSMParameterReadOnlyWithHistory. The default read-only set permits DescribeParameters, GetParameter, and GetParameters but omits GetParameterHistory.

Can a CloudFormation template use a shared Parameter Store value through a dynamic reference?

No. Shared parameters support CloudFormation template parameter inputs, but not CloudFormation dynamic references.

A team receives a Parameter Store share from outside its AWS organization but cannot access the parameter yet. What sharing step may be missing?

Accept the AWS RAM resource-share invitation. Invitation-free access requires enabled sharing within the same organization.

Can a recipient update or reshare a Parameter Store parameter received through AWS RAM, and what happens when the owner removes the share?

No. Only the owner can change or share it. Removing the parameter from the share ends recipient access.

Which account pays for a shared advanced Parameter Store value and which account consumes its read throughput?

The owner pays storage. Each consumer pays its own API usage and has retrieval throughput enforced separately from other consumers.

Can a Systems Manager Parameter Store advanced parameter be changed to standard after its value becomes smaller?

No. Standard can be upgraded to advanced, but an advanced parameter cannot be downgraded.

Which Systems Manager Parameter Store policies exist, and how do they notify?

Expiration deletes the parameter; ExpirationNotification and NoChangeNotification emit EventBridge events. Advanced parameters support up to ten policies; a newly supplied policy array replaces the existing policies.

An application needs all Systems Manager parameters below its environment path. Which API and naming structure fit?

Use slash-separated parameter hierarchies and GetParametersByPath with recursive retrieval. Hierarchies support up to 15 levels.

A Systems Manager parameter already has 100 versions and the oldest has a deployment label. What happens when its value is updated?

The update fails because the labeled oldest version cannot be deleted. Move the label to a newer version before continuing updates.

Engineers should run only an approved interactive command through Systems Manager. How can access be restricted?

Create a custom Session document with schemaVersion 1.0 and sessionType InteractiveCommands, then restrict IAM access to that document. These sessions must start through the CLI.

A Systems Manager Run As session fails because its configured Linux user is absent. Will it fall back to ssm-user?

No. The configured user must exist; Run As fails without falling back to ssm-user.

What is the default idle timeout for Systems Manager Session Manager, and does it impose a strict total duration?

20 minutes, configurable from one to 60 minutes. Client inputs such as resizing or reconnection reset it; use maximum session duration for a separate total-duration control.

A team wants native Systems Manager sessions on Linux to use an existing operating-system user rather than ssm-user. What should it configure?

Enable Run As. Set SSMSessionRunAs on the IAM user or role, or set the OS user in session preferences. The IAM tag takes precedence, and the selected user must exist.

A Systems Manager maintenance-window task fails. Does task priority automatically skip all later tasks?

No. Later-priority tasks can start after earlier tasks reach a final state, regardless of success. Priority orders tasks; it is not a success dependency.

A Systems Manager maintenance window reaches its cutoff while tasks are running. What happens by default?

CONTINUE_TASK allows running tasks to continue. Cutoff stops scheduling new tasks; it is not an automatic termination deadline.

Which Step Functions workflow type can run as a Systems Manager maintenance-window task?

Standard workflows. Express workflows are unsupported.

Você concluiu todos os tópicos da DOP-C02!

Seu badge ficou dourado. Próximo passo: agendar a prova na AWS. Depois de passar, adicione-se ao Quem passou.

Compartilhar

Conte que você concluiu esta certificação

Concluí todos os tópicos da AWS Certified DevOps Engineer - Professional (DOP-C02) no Easy Review. Venha estudar também!