Open efb4f5ff-1298-471a-8973-3d47447115dc opened 1 year ago
Hello, @efb4f5ff-1298-471a-8973-3d47447115dc ! Thank you for your input, we'll take a look at the issue.
Hello @efb4f5ff-1298-471a-8973-3d47447115dc ,
From the URL you have sent everything looks working as expected: May 8, the issue was labeled as staled because of 28 days of inactivity - as it should be actions added the comment about the issue to be closed in 7 days May 15, the issue is closed - as it should be
Also it is not possible to reproduce the problem:
The test build runs agains the test issue and it closes it.
If you think the action does not work properly can you please provide the exact configuration of the build: its workflow YAML file and if possible - logs of the stale build with pointed
I think u overlooked a few things in the issue
From the URL you have sent everything looks working as expected: May 8, the issue was labeled as staled because of 28 days of inactivity - as it should be actions added the comment about the issue to be closed in 7 days
This is true but look at the picture i added below, the label was added on 8 may but also removed by the bot on 8 may
May 15, the issue is closed - as it should be
The bot did not close the issue. I did that myself
This was the config at the time and hasnt changed since: https://github.com/FreeTubeApp/FreeTube/blob/development/.github/workflows/stale.yml
logs: https://github.com/FreeTubeApp/FreeTube/actions/runs/4910543174/job/13300466438
Thank you @efb4f5ff-1298-471a-8973-3d47447115dc for the config. It helps a lot to reproduce the problem but all the runs still works as expected.
Nevertheless it was discovered the issue was un-staled on the next run May 10
Please give us some more time to investigate why it could happen
Hello @efb4f5ff-1298-471a-8973-3d47447115dc , the only possible reason the issue to be un-staled is the false detection of update caused by the delay between creating the label and marking the issue as updated. The action has no control on infrastructure and works it out by having the tolerance of 15 secs.
It seems in the case of the issue the delay was more than 15 sec. It is not common situation and we will gather more info before taking any action.
Hi @dsame, I'm seeing this same issue fairly consistently. I'm running v8.0.0 in debug-only
mode and all 12 of these "No longer stale PRs" were caused by the tolerance exceeding 15 secs (i.e. difference between the stale
label event's created_at
and the PR's updated_at
):
Statistics:
Processed PRs: 2326
No longer stale PRs: 12
Deleted PRs labels: 12
Fetched items: 2400
Fetched items events: 33
Fetched items comments: 33
Operations performed: 102
It was as high as 49sec in some cases. Happy to provide any more details if you need any clarification. Possibly relevant: we're using a monorepo, and this has led to more exaggerated "eventual consistency" issues due to higher volumes of activity.
A few potential ideas:
Thanks!
Experiencing this as well, is there a workaround for this?
Seems like we've just run into this too, it keeps adding and deleting the stale
label without actually closing the PR
https://github.com/nginxinc/nginx-gateway-fabric/pull/1623#issuecomment-2028533314
Experiencing this as well, is there a workaround for this?
The workaround I've been using is to set remove-stale-when-updated
to false
. After the issue described here is fixed, I plan on switching back to the default true
.
Description: Bot removes label
Action version: v8
Platform:
Runner type:
Repro steps:
In https://github.com/FreeTubeApp/FreeTube/issues/754#issuecomment-1537626941 u can see that the bot should close the issue but removed the label instead
Expected behavior: Should close the issue
Actual behavior: Removes the label preventing it from closing the issue