This message was deleted.
# atlantis-community
s
This message was deleted.
p
even if you delete the data dir?
like on a clean install?
w
this was a PR that wasn’t planned before the restart, and IIRC, I don’t have it setup to use a persistent volume.
so I think the behavior would be the same either way
p
but why atlantis will look for the dir of the PR checkout if there was no event?
how atlantis will know the existence of the PR at that point?
w
I’m not sure. all I know is
atlantis plan -d xxx
— didn’t work
atlantis plan
- did work
the last commit to that PR was likely before Atlantis was restarted (and we currently don’t have a persistent volume setup for the atlantis user’s directory, so it’s wiped when the set restarts)
p
that si very strange
maybe during the restart you had a webhook trying to reach?
w
there should not have been a data dir there at the time I ran that command as best I know, which is why the error was confusing. if I did a regular
atlantis plan
it worked fine, and there were no restarts in between
p
that is very strange
j
I saw something similar right now. We kept seeing this occasionally after pod restarts. 1. Created PR, nothing was planned because autoplan didn’t catch the changes 2. Pod was restarted 3. Tried to
atlantis plan -p <project>
a. Failed with
checking if workspace exists: stat
above 4. Tried
atlantis plan
- nothing was planned (due to bad
autoplan
I assume) 5. Tried
atlantis plan -p .*
a. That actually started planning correctly
j
Has a bug been filed for this issue? I’ve been seeing this as well.
p
and this was not happening in earlier versions?
0.22.x?
j
I need to go check our archives, but this has actually been happening to us for quite a while now. Like almost since we started using Atlantis a year ago.
p
@wby did you created an issue for this? maybe you guys can create the issue and add debug logs and such. If the issue is old then if any of you is willing to put the time to investigate and create a PR that will be great
w
I never created an issue for it. not sure I have debug logs that far back. so far, we only saw it when atlantis was restarted in the middle of a PR being open
j
Getting back to this. I wanted to do some testing before commenting back. We are seeing this issue regardless of whether or not Atlantis has been restarted or not. If I open a brand new PR, and plan a single workspace it fails with a
no such file or directory
error. We do not have autoplanning enabled fwiw. If I do an
atlantis plan
things work as expected.
In this particular case, I’ve only modified a single workspace, so
atlantis plan -p <workspace>
is planning the same project as
atlantis plan
p
and you have a custom atlantis.yaml?
j
Yes I do
p
I think we need an issue for this, the slack history gets lost
👍 1
do you think you could create an issue and post the debug logs , atlantis.yaml and repos.yaml , atlantis version etc?
that will be very useful
p
Is an issue available for this? I’m seeing the same after going from
0.22.3
to
0.23.4
, we also have a custom workflow.
p
I will say wait for 0.23.5 and then test it out, i should be out in a few minutes
p
thankyou
p
could you try with the new release @Phil Gardner?
p
Just saw this, GitHub is having another incident right now around Webhooks so I’ll have to try later.