-
Notifications
You must be signed in to change notification settings - Fork 265
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Regression after recent commit, Segmentation fault #621
Comments
I know this is not an error on an active release... but seems a bit important to ignore... I don't want to be that guy.. but is this project actively maintained? |
I can confirm that this issue is still happening. |
After some investigation I think the issue is related to the version bump of github.com/argoproj/argo-cd/v2 v2.2.7 to version v2.7.7 |
I found the issue, it's here :
Starting with the MR 8516 linked above, we need a CredsStore to fetch credentials. If we pass nil we segfault because it's not what's excepted. I tried to pass a repo.GetGitCreds(git.NoopCredsStore{}) but I end up with the following stack trace:
Which is a bit better but still not working. I should mention that my credentials are effectively present in the created Git client (not an issue with my credentials per se). To be honest I am not really sure what is supposed to be passed in the credsStore, @alexmt, @jannfis I saw that you participated in the MR that introduced this change (argoproj/argo-cd#8516). Any help on the subject ? |
Updated to the new 0.13 when it came out a couple of days ago, and haven't seen any updates since. Just noticed that it's because of this bug, the container just crashes as soon as an update is found. |
It's happening to me too in the new version v0.13. With old version v0.12.2 works all OK. Any help? Thanks. |
Same here. What's strange: I'm operating two clusters (for dev and prod) and in the dev cluster it's running. Both clusters have an equivalent configuration, i.e. the config differs mostly in secrets (and scalability ofc). So why it's running fine in my dev cluster, but in my prod cluster it's bailing out?! |
Here's the stack trace:
Looking at the source code, it seems like |
I'll take this recent activity on this issue to mention that the delay in handling this (even after my very detailed report), or even getting any feedback was one of the main reasons for our team choosing to not go for this tool. Seeing as there even was a new release without even addressing this at all I feel justified in our decision. Sorry for minor venting |
I did an experiment and changed my configuration from HTTPSCreds to GitHubAppCreds, but the behavior is the same:
When you check the first stack frame you can see that it now bails out in line 388 instead of line 218 in the exact same file |
@ImFlog I am trying this:
with
|
Describe the bug
Segmentation fault due to (semi-)recent commit.
On release v0.12.2 argocd-image-updater functions properly. But the current version in the master branch segfaults.
The commit to blame is most probably d5a8f94 , this is the first one to break from the ones between v0.12.2 and now.
To Reproduce
Not extremely straightforward, but currently using a very simple setup, with a helm chart for argocd-image-updater.
Chart config:
And the relevant argo-Application annotations
Expected behavior
To not segfault :)
Additional context
Encountered while trying to add a feature required for my team
Version
First commit to break seems to be d5a8f94
Logs
Crash logs
The text was updated successfully, but these errors were encountered: