Improve export logs user experience

[hgreebe]

Description of changes

• Add the following improvements to export-logs:
  • Progress reporting: During the export, add await poll that emits a line to stderr every 15s.
  • Parallel download: There is no native boto3 "download folder" boto3 API. Therefore, we have to use a Python thread pool over download_file boto3 API. Objects are grouped to be handled sequentially in different workers.
  • Track an existing running CloudWatch log export task: If the running task matches the same log group, track it to completion, instead of failing. A different log group's task yields a clear "please wait" failure message.
  • Support export logs from deleted clusters: To export logs from a deleted cluster, ARN of the CloudFormation needs to be provided.
• Added unit tests to cover improvements

Tests

• Ran export logs in multiple scenarios (output added below)
• Added unit tests

Output when running export logs:

pcluster export-cluster-logs --cluster-name {name}
Exporting 10326 log stream(s) from log group /aws/parallelcluster/{name}...
Started CloudWatch Logs export task XXXX-XXXX. Waiting for it to complete... (CloudWatch buffers logs and usually writes them to S3 near the end of the task, so the exported size may stay at 0 for several minutes.)
Exporting logs from CloudWatch: no data written to S3 yet (elapsed 15s)
...
Exporting logs from CloudWatch: 85.3 MB exported so far (elapsed 124s)
Export task XXXX-XXXX finished after 138s.
Downloading 9387 log object(s) (106.5 MB) from S3...
...
Downloaded 8819/9387 log object(s)... (elapsed 346s)
Downloaded 9387 of 9387 log object(s) (106.5 MB). (elapsed 357s)
CloudWatch logs export complete.
{
  "url": "{url}"
}

Output when another export on different cluster is running:

pcluster export-cluster-logs --cluster-name {name}
Exporting 10 log stream(s) from log group /aws/parallelcluster/{name}
ERROR: Unable to export cluster's logs.
Unexpected error when exporting cluster's logs: Another CloudWatch Logs export task is already running for this account (task XXXX-XXXX, log group '/aws/parallelcluster/{other_cluster}', started at 2026-08-12 15:38:20.354000+00:00). CloudWatch Logs allows only one export task at a time; please wait for it to complete before retrying.

Output when another export is running on the same cluster:

pcluster export-cluster-logs --cluster-name {name}
Exporting 10326 log stream(s) from log group /aws/parallelcluster/{name}...
An export task for log group '/aws/parallelcluster/{name}' is already running (task XXXX-XXXX, started at 2026-08-12 15:38:20.354000+00:00). Tracking it instead of starting a new one.
Started CloudWatch Logs export task XXXX-XXXX. Waiting for it to complete... (CloudWatch buffers logs and usually writes them to S3 near the end of the task, so the exported size may stay at 0 for several minutes.)
...

References

• Based off of: hanwen-cluster@27ed5d8

Checklist

• Make sure you are pointing to the right branch.
• If you're creating a patch for a branch other than develop add the branch name as prefix in the PR title (e.g. [release-3.6]).
• Check all commits' messages are clear, describing what and why vs how.
• Make sure to have added unit tests or integration tests to cover the new/modified code.
• Check if documentation is impacted by this change.

Please review the guidelines for contributing and Pull Request Instructions.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[hanwen-cluster]

We need to use AWSApi.instance().s3.download_file instead of s3_resource because only boto client is thread safe: https://docs.aws.amazon.com/boto3/latest/guide/clients.html:

Multithreading or multiprocessing with clients
Unlike Resources and Sessions, clients are generally thread-safe. There are some caveats, defined below, to be aware of though.

Mistakenly approved