gh-140279: Stale workflow needs 'actions: write' to update its own cache

[hugovk]

Looking at the last three stale runs:

• https://github.com/python/cpython/actions/runs/24013745584/job/70029433814
• https://github.com/python/cpython/actions/runs/24021719127/job/70051782660
• https://github.com/python/cpython/actions/runs/24031392929/job/70080736928

They all process the same batch:

Processing the batch of issues #1 containing 100 issues...
::group::[#37883] Issue #37883
[#37883] Issue #37883
##[debug]state: mark 37883 as processed
...
::group::[#13536] Pull request #13536
[#13536] Pull request #13536
##[debug]state: mark 13536 as processed
Warning: No more operations left! Exiting...
Warning: If you think that not enough issues were processed you could try to increase the quantity related to the operations-per-run (​[https://github.com/actions/stale#operations-per-run​)](https://github.com/actions/stale#operations-per-run%E2%80%8B)) option which is currently set to 120
Statistics:
Processed items: 2498
├── Processed issues: 2446
└── Processed PRs : 52
Fetched items: 2600
Fetched items events: 47
Fetched items comments: 47
Operations performed: 120

This is because it fails to update its own cache:

state: persisting info about 2561 issue(s)
##[debug]remove cache "_state"
Warning: Error delete _state: [403] Resource not accessible by integration - https://docs.github.com/rest/actions/cache#delete-github-actions-caches-for-a-repository-using-a-cache-key
##[debug]Cache service version: v2
##[debug]Checking zstd --quiet --version
##[debug]1.5.7
##[debug]zstd version: 1.5.7
##[debug]implicitDescendants 'false'
##[debug]followSymbolicLinks 'true'
##[debug]implicitDescendants 'false'
##[debug]matchDirectories 'true'
##[debug]omitBrokenSymbolicLinks 'true'
##[debug]excludeHiddenFiles 'false'
##[debug]Search path '/tmp/56acbeaa-1fef-4c79-8f84-7565e560fb03'
##[debug]Matched: ../../../../../tmp/56acbeaa-1fef-4c79-8f84-7565e560fb03
##[debug]Cache Paths:
##[debug]["../../../../../tmp/56acbeaa-1fef-4c79-8f84-7565e560fb03"]
##[debug]Archive Path: /home/runner/work/_temp/182ce2b9-beec-48f1-8a39-01885e86ce30/cache.tzst
/usr/bin/tar --posix -cf cache.tzst --exclude cache.tzst -P -C /home/runner/work/cpython/cpython --files-from manifest.txt --use-compress-program zstdmt
/usr/bin/tar -tf /home/runner/work/_temp/182ce2b9-beec-48f1-8a39-01885e86ce30/cache.tzst -P --use-compress-program unzstd
../../../../../tmp/56acbeaa-1fef-4c79-8f84-7565e560fb03/
../../../../../tmp/56acbeaa-1fef-4c79-8f84-7565e560fb03/state.txt
##[debug]File Size: 5712
##[debug]Reserving Cache
##[debug][Request] CreateCacheEntry https://results-receiver.actions.githubusercontent.com/twirp/github.actions.results.api.v1.CacheService/CreateCacheEntry
##[debug][Response] - 409
##[debug]Headers: {
##[debug] "content-length": "98",
##[debug] "content-type": "application/json",
##[debug] "date": "Mon, 06 Apr 2026 12:15:29 GMT",
##[debug] "x-github-backend": "Kubernetes",
##[debug] "x-github-request-id": "C440:1E9CFA:222389D:28AADDD:69D3A3E0"
##[debug]}
##[debug]Body: {
##[debug] "code": "already_exists",
##[debug] "msg": "cache entry with the same key, version, and scope already exists"
##[debug]}
##[debug]Failed to reserve cache: Error: Failed to CreateCacheEntry: Received non-retryable error: Failed request: (409) Conflict: cache entry with the same key, version, and scope already exists
Failed to save: Unable to reserve cache with key _state, another job may be creating this cache.

According to actions/stale#1133 (comment) it needs actions: write to update the cache.

Then the next run should be able to process the next batch.

• Issue: Mark stale prs CI has a very low limit #140279

[StanFromIreland]

You can put this under #140279 (which seeks to improve its effectiveness).

[hugovk]

The newest one had no cache error 👍

• https://github.com/python/cpython/actions/runs/24044331914/job/70123774859

[hugovk]

And the next two have continued as expected:

• https://github.com/python/cpython/actions/runs/24057861683/job/70167689553
• https://github.com/python/cpython/actions/runs/24067726705/job/70197287785

[Safihre]

So, what is actually done with the Stale label? Does anyone look at them?
Since there is no auto-closing?

[hugovk]

See #116857 (comment), let's continue in core-workflow.