Understanding command statuses - AWS Systems Manager

Understanding command statuses

Run Command, a capability of AWS Systems Manager, reports detailed status information about the different states a command experiences during processing and for each managed node that processed the command. You can monitor command statuses using the following methods:

  • Choose the Refresh icon on the Commands tab in the Run Command console interface.

  • Call list-commands or list-command-invocations using the AWS Command Line Interface (AWS CLI). Or call Get-SSMCommand or Get-SSMCommandInvocation using AWS Tools for Windows PowerShell.

  • Configure Amazon EventBridge to respond to state or status changes.

  • Configure Amazon Simple Notification Service (Amazon SNS) to send notifications for all status changes or specific statuses such as Failed or TimedOut.

Run Command status

Run Command reports status details for three areas: plugins, invocations, and an overall command status. A plugin is a code-execution block that is defined in your command's SSM document. For more information about plugins, see Command document plugin reference.

When you send a command to multiple managed nodes at the same time, each copy of the command targeting each node is a command invocation. For example, if you use the AWS-RunShellScript document and send an ifconfig command to 20 Linux instances, that command has 20 invocations. Each command invocation individually reports status. The plugins for a given command invocation individually report status as well.

Lastly, Run Command includes an aggregated command status for all plugins and invocations. The aggregated command status can be different than the status reported by plugins or invocations, as noted in the following tables.

Note

If you run commands to large numbers of managed nodes using the max-concurrency or max-errors parameters, command status reflects the limits imposed by those parameters, as described in the following tables. For more information about these parameters, see Run commands at scale.

Detailed status for command plugins and invocations
Status Details
Pending The command hasn't yet been sent to the managed node or hasn't been received by SSM Agent. If the command isn't received by the agent before the length of time passes that is equal to the sum of the Timeout (seconds) parameter and the Execution timeout parameter, the status changes to Delivery Timed Out.
InProgress Systems Manager is attempting to send the command to the managed node, or the command was received by SSM Agent and has started running on the instance. Depending on the result of all command plugins, the status changes to Success, Failed, Delivery Timed Out, or Execution Timed Out. Exception: If the agent isn't running or available on the node, the command status remains at In Progress until the agent is available again, or until the execution timeout limit is reached. The status then changes to a terminal state.
Delayed The system attempted to send the command to the managed node but wasn't successful. The system retries again.
Success This status is returned under a variety of conditions. This status doesn't mean the command was processed on the node. For example, the command can be received by SSM Agent on the managed node and return an exit code of zero as a result of your PowerShell ExecutionPolicy preventing the command from running. This is a terminal state. Conditions that result in a command returning a Success status are:
  • When targeting a single instance, the command was received by SSM Agent on the managed node and returned an exit code of zero.

  • When targeting multiple instances, the number of failed invocations has not crossed the error threshold specified in the command.

  • When targeting multiple instances, at least 1 invocation has succeeded while others have timed out. The specified error threshold still applies.

  • When targeting a tag, no instances are found associated with the tag.

  • When targeting a tag, the number of failed invocations has not crossed the error threshold specified in the command.

  • When targeting a tag, at least 1 invocation has succeeded while others have timed out. The specified error threshold still applies.

  • You have applications or policies enforced at the OS level that prevent or override the execution of a command resulting in an exit code of zero being returned.

Note

The same conditions apply when targeting resource groups. To troubleshoot errors or get more information about the command execution, send a command that handles errors or exceptions by returning appropriate exit codes (non-zero exit codes for command failure).

DeliveryTimedOut The command wasn't delivered to the managed node before the total timeout expired. Total timeouts don't count against the parent command’s max-errors limit, but they do contribute to whether the parent command status is Success, Incomplete, or Delivery Timed Out. This is a terminal state.
ExecutionTimedOut Command automation started on the managed node, but the command wasn’t completed before the execution timeout expired. Execution timeouts count as a failure, which will send a non-zero reply and Systems Manager will exit the attempt to run the command automation, and report a failure status.
Failed The command wasn't successful on the managed node. For a plugin, this indicates that the result code wasn't zero. For a command invocation, this indicates that the result code for one or more plugins wasn't zero. Invocation failures count against the max-errors limit of the parent command. This is a terminal state.
Cancelled The command was canceled before it was completed. This is a terminal state.
Undeliverable The command can't be delivered to the managed node. The node might not exist or it might not be responding. Undeliverable invocations don't count against the parent command’s max-errors limit, but they do contribute to whether the parent command status is Success or Incomplete. For example, if all invocations in a command have the status Undeliverable, then the command status returned is Failed. However, if a command has five invocations, four of which return the status Undeliverable and one of which returns the status Success, then the parent command's status is Success. This is a terminal state.
Terminated The parent command exceeded its max-errors limit and subsequent command invocations were canceled by the system. This is a terminal state.
InvalidPlatform The command was sent to a managed node that didn't match the required platforms specified by the chosen document. Invalid Platform doesn't count against the parent command’s max-errors limit, but it does contribute to whether the parent command status is Success or Failed. For example, if all invocations in a command have the status Invalid Platform, then the command status returned is Failed. However, if a command has five invocations, four of which return the status Invalid Platform and one of which returns the status Success, then the parent command's status is Success. This is a terminal state.
AccessDenied The AWS Identity and Access Management (IAM) user or role initiating the command doesn't have access to the targeted managed node. Access Denied doesn't count against the parent command’s max-errors limit, but it does contribute to whether the parent command status is Success or Failed. For example, if all invocations in a command have the status Access Denied, then the command status returned is Failed. However, if a command has five invocations, four of which return the status Access Denied and one of which returns the status Success, then the parent command's status is Success. This is a terminal state.
Detailed status for a command
Status Details
Pending The command wasn't yet received by an agent on any managed nodes.
InProgress The command has been sent to at least one managed node but hasn't reached a final state on all nodes.
Delayed The system attempted to send the command to the node but wasn't successful. The system retries again.
Success The command was received by SSM Agent on all specified or targeted managed nodes and returned an exit code of zero. All command invocations have reached a terminal state, and the value of max-errors wasn't reached. This status doesn't mean the command was successfully processed on all specified or targeted managed nodes. This is a terminal state.
Note

To troubleshoot errors or get more information about the command execution, send a command that handles errors or exceptions by returning appropriate exit codes (non-zero exit codes for command failure).

DeliveryTimedOut The command wasn't delivered to the managed node before the total timeout expired. The value of max-errors or more command invocations shows a status of Delivery Timed Out. This is a terminal state.
Failed

The command wasn't successful on the managed node. The value of max-errors or more command invocations shows a status of Failed. This is a terminal state.

Incomplete The command was attempted on all managed nodes and one or more of the invocations doesn't have a value of Success. However, not enough invocations failed for the status to be Failed. This is a terminal state.
Cancelled The command was canceled before it was completed. This is a terminal state.
RateExceeded The number of managed nodes targeted by the command exceeded the account quota for pending invocations. The system has canceled the command before executing it on any node. This is a terminal state.
AccessDenied The user or role initiating the command doesn't have access to the targeted resource group. AccessDenied doesn't count against the parent command’s max-errors limit, but does contribute to whether the parent command status is Success or Failed. (For example, if all invocations in a command have the status AccessDenied, then the command status returned is Failed. However, if a command has 5 invocations, 4 of which return the status AccessDenied and 1 of which returns the status Success, then the parent command's status is Success.) This is a terminal state.
No Instances In Tag The tag key-pair value or resource group targeted by the command doesn't match any managed nodes. This is a terminal state.

Understanding command timeout values

Systems Manager enforces the following timeout values when running commands.

Total Timeout

In the Systems Manager console, you specify the timeout value in the Timeout (seconds) field. After a command is sent, Run Command checks whether the command has expired or not. If a command reaches the command expiration limit (total timeout), it changes status to DeliveryTimedOut for all invocations that have the status InProgress, Pending or Delayed.

The Timeout (seconds) field in the Systems Manager console

On a more technical level, total timeout (Timeout (seconds)) is a combination of two timeout values, as shown here:

Total timeout = "Timeout(seconds)" from the console + "timeoutSeconds": "{{ executionTimeout }}" from your SSM document

For example, the default value of Timeout (seconds) in the Systems Manager console is 600 seconds. If you run a command by using the AWS-RunShellScript SSM document, the default value of "timeoutSeconds": "{{ executionTimeout }}" is 3600 seconds, as shown in the following document sample:

"executionTimeout": { "type": "String", "default": "3600", "runtimeConfig": { "aws:runShellScript": { "properties": [ { "timeoutSeconds": "{{ executionTimeout }}"

This means the command runs for 4,200 seconds (70 minutes) before the system sets the command status to DeliveryTimedOut.

Execution Timeout

In the Systems Manager console, you specify the execution timeout value in the Execution Timeout field, if available. Not all SSM documents require that you specify an execution timeout. The Execution Timeout field is only displayed when a corresponding input parameter has been defined in the SSM document. If specified, the command must complete within this time period.

Note

Run Command relies on the SSM Agent document terminal response to determine whether or not the command was delivered to the agent. SSM Agent must send an ExecutionTimedOut signal for an invocation or command to be marked as ExecutionTimedOut.

The Execution Timeout field in the Systems Manager console
Default Execution Timeout

If a SSM document doesn't require that you explicitly specify an execution timeout value, then Systems Manager enforces the hard-coded default execution timeout.

How Systems Manager reports timeouts

If Systems Manager receives an execution timeout reply from SSM Agent on a target, then Systems Manager marks the command invocation as executionTimeout.

If Run Command doesn't receive a document terminal response from SSM Agent, the command invocation is marked as deliveryTimeout.

To determine timeout status on a target, SSM Agent combines all parameters and the content of the SSM document to calculate for executionTimeout. When SSM Agent determines that a command has timed out, it sends executionTimeout to the service.

The default for Timeout (seconds) is 3600 seconds. The default for Execution Timeout is also 3600 seconds. Therefore, the total default timeout for a command is 7200 seconds.

Note

SSM Agent processes executionTimeout differently depending on the type of SSM document and the document version.