-
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
Bazel hangs forever #4322
Comments
Thanks for you bug report and the stack trace! As it hangs waiting for a lock, we might have to reconsider the locking strategy in use (I didn't find an obvious lock order violation; it seems that we only every ask for the parent lock while holding a lock on a child directory). /cc @tomlu who last changed the locking strategy for that part of the code in 82e68b7. This may or may not be related to #4306. |
I can 100% reproduce this bug. I can provide more information if you need. |
@snnn does it happen more frequently if you increase --jobs (parallelism)? In the code we use guava's
The locks are guaranteed to be re-entrant, but it's possible that on two different threads createDirectory gets two locks in reverse order i.e. The number of stripes is set to 64, so if you run with lots of jobs for a while it may not be that unlikely to get into that state. https://google.github.io/guava/releases/19.0/api/docs/com/google/common/util/concurrent/Striped.html |
Sorry about this and thanks for the expedient rollback. This is the second
time I've broken this bit of the code. I'll add a regression test before I
touch it again.
…On Tue, Dec 19, 2017 at 10:14 AM, bazel.io machine account < ***@***.***> wrote:
Closed #4322 <#4322> via c801c39
<c801c39>
.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#4322 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/ABbnSmaWxoxjcVXcIJgusfgIUCKhwaxrks5tB9LPgaJpZM4RGSQe>
.
|
Fixes #4322, #4306. *** Reason for rollback *** Introduces a deadlock (see bazelbuild/bazel#4322) *** Original change description *** Make FileSystem operate on LocalPath instead of Path. PiperOrigin-RevId: 179549866
Description of the problem / feature request / question:
While running tensorflow bazel CI build on Windows, it hangs forever.
If possible, provide a minimal example to reproduce the problem:
I'm compiling meteorcloudy's TF code, with some private changes
meteorcloudy/tensorflow@7004ed6
Environment info
Operating System:
Windows 2012
Bazel version (output of
bazel info release
):master
If
bazel info release
returns "development version" or "(@non-git)", please tell us what source tree you compiled Bazel from; git commit hash is appreciated (git rev-parse HEAD
):1892677
Have you found anything relevant by searching the web?
No
Anything else, information or logs or outputs that would be helpful?
The text was updated successfully, but these errors were encountered: