-
Notifications
You must be signed in to change notification settings - Fork 4k
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
Windows repository_ctx.execute(command)
sometimes returns incomplete result
#2675
Comments
Any progress? This is killing my presubmits. |
Still haven't found the root cause, but looks like there is already a workaround for the Kokoro presubmit?(cl/150360798) I'll keep digging... |
This issue might have something to do with #2774 |
@laszlocsomor Can you check if there is similar problem in SkylarkExecutionResult.java ? |
Dupe for #2774 ? |
@dslomov : I'm not convinced of that yet. Are you sure? |
Missed a question mark :) |
Ah :) I'm trying to repro it now. |
@meteorcloudy : re #2675 (comment): I don't think so, I believe we may be losing the stdout/stderr somehow. Since you were unable to repro this locally, I thought maybe because you didn't redirect the output to a file, whereas on CI it is redirected I believe. Anyway, I set up an experiment with repo rules: I was building two of them in parallel, a fast and a slow one. It seems repo rules' actions ( @dslomov : because of the reasons above, I doubt this bug is a dupe of #2774. |
Is this still expected to be fixed in 0.5? We still have some release blockers and we will not cut the release sooner than in 2 weeks, but we should be getting ready to. |
I do not think we have a repro, so we do not know if it is fixed or not. I do not think we need to treat it as a release blocker. |
I've added stricter error checks and more logging to |
Ok thanks, not treating as release blocker. Good luck! :) |
Let's move it to 0.6 |
Add stricter error checks in hopes of catching occasional CI flakiness where the stdout of a command seems to get lost. It's now an error if the command returns a non-zero exit code (or a zero one if it's expected to fail) or if its stdout is empty. Previously we only checked if stderr was empty to consider the action successful. See #2675 RELNOTES: none PiperOrigin-RevId: 152685220
I can reproduce it on 688dbf7 in a CI system. As @meteorcloudy , I could never reproduce this on my local machine. |
I finally figure out the root cause, it's |
Hi @meteorcloudy , How is it going? Is it solved? |
Well, it turned out to be much more complicated that I thought. But yes! I have a fix under review, it will be fixed soon. https://bazel-review.googlesource.com/#/c/bazel/+/18790/ |
This is causing #2434 and the internal Kokoro Windows Flakiness: missing "LIB"
The text was updated successfully, but these errors were encountered: