-
Notifications
You must be signed in to change notification settings - Fork 3.2k
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
Intermittent segmentation faults from rustc on github_arm_linux_runner_2404 #11533
Comments
This issue may be related though the failures I see are not exactly the same: rust-lang/rust#125430 |
We run the same workflow on a non-arm runner and have not seen these failures. |
I have not had much luck reproducing locally but I was able to reproduce on the repo in question with a very minimal workflow. This same workflow does not reproduce the issue on my public repo: workflow:
log:
|
Here's a log from the public repo.
|
Hi @westonpace, Thank you for bringing this issue to our attention. We are looking into this issue and will update you on this issue after investigating. |
Interesting, same timeframe when this started for us but impacting a terragrunt based pipeline running on GH Large Hosted Arm64 runners 😢 |
@westonpace , which public repo are you referring to in this comment? We would like to get a solid repro we can work with. |
@tedchamb sorry if I was unclear. I cannot reproduce on a public repo. Here was my attempt: https://github.com/lancedb/rustc-error-repro If I run this exact same workflow in my private repo then I get the error. It does not occur on a public repo. |
This issue is related to ARM image for Github actions. |
Thank you for sharing that repository. I have found actions/partner-runner-images#40 and added some comments there. |
. |
This is being tracked in the partner repo. |
Description
The
rustc
on the runner image will sometimes abort with a segmentation fault. I have even seen this happen when just usingrustc
to print the current version ofrustc
.Platforms affected
Runner images affected
Image version and build link
Here is the log of a failed run:
Is it regression?
Yes, we did not see this on 2.321.0
Expected behavior
I would expect I can either use the version of rustc that is on the runner image or I can download, install, and use a clean version of rustc using rustup.
For reference, here is a log from a successful build (truncated after the step that is failing in 2.322.0) from 2.321.0:
Actual behavior
The failure is not consistent. Slightly less than half of our builds succeed. The builds that fail will sometimes fail during the rustup stage, sometimes after that when printing the current rustc version, and sometimes will fail during the compilation itself. Everytime it fails it is some strange segmentation fault. I suspect the rustc on the image is corrupt somehow.
The first failure appeared on January 29th, around 9:30 AM PST.
We sometimes get failures even when we try and use rustup to install an older version of rustc. I believe this is because the first thing that rustup does is list the current rust version and this can raise a segmentation fault.
Repro steps
The repository I am encountering this in is a private repository but I can share the workflow. In this workflow and the example logs I have replaced the repository name with
redacted
. I will attempt to reproduce this on an open repository:The text was updated successfully, but these errors were encountered: