Viewing Pipeline Logs

Pipeline-level logging is supported at pipeline creation by specifying an Amazon S3 location in either the console or with a pipelineLogUri in the default object in SDK/CLI. The directory structure for each pipeline within that URI is like the following:

pipelineId
    -componentName
        -instanceId
            -attemptId

For pipeline, df-00123456ABC7DEF8HIJK, the directory structure looks like:

df-00123456ABC7DEF8HIJK
    -ActivityId_fXNzc
        -@ActivityId_fXNzc_2014-05-01T00:00:00
            -@ActivityId_fXNzc_2014-05-01T00:00:00_Attempt=1

For ShellCommandActivity, logs for stderr and stdout associated with these activities are stored in the directory for each attempt.

For resources like, EmrCluster, where an emrLogUri is set, that value takes precedence. Otherwise, resources (including TaskRunner logs for those resources) follow the above pipeline logging structure.

To view logs for a given pipeline run:

1. Retrieve the ObjectId by calling query-objects to get the exact object ID. For example:

   aws datapipeline query-objects --pipeline-id <pipeline-id> --sphere ATTEMPT --region ap-northeast-1

   query-objects is a paginated CLI and may return a pagination token if there are more executions for the given pipeline-id. You can use the token to go through all the attempts until you find the expected object. For example, a returned ObjectId would look like: @TableBackupActivity_2023-05-020T18:05:18_Attempt=1.

2. Using the ObjectId, retrieve the log location using:

   aws datapipeline describe-objects —pipeline-id <pipeline-id> --object-ids <object-id> --query "pipelineObjects[].fields[?key=='@logLocation'].stringValue"

Error message of a failed activity

To get the error message, first get the ObjectId using query-objects.

After retrieving the failed ObjectId, use the describe-objects CLI to get the actual error message.

aws datapipeline describe-objects --region ap-northeast-1 --pipeline-id <pipeline-id> --object-ids <object-id> --query "pipelineObjects[].fields[?key=='errorMessage'].stringValue"

Cancel or rerun or mark as finished an object

Use the set-status CLI to cancel a running object, or re-run a failed object or mark a running object as Finished.

First, get the object ID using the query-objects CLI. For example:

aws datapipeline query-objects --pipeline-id <pipeline-id> --sphere INSTANCE --region ap-northeast-1

Use the set-status CLI to change the status of the desired object. For example:

aws datapipeline set-status —pipeline-id <pipeline-id> --region ap-northeast-1 --status TRY_CANCEL --object-ids <object-id>