-
Notifications
You must be signed in to change notification settings - Fork 12.7k
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
clangd/unittests/TUSchedulerTests.cpp fails and timesout frequently #64964
Comments
@llvm/issue-subscribers-bug |
@llvm/issue-subscribers-clangd |
Created a patch to configure clangd tests apart from clang tools tests. https://reviews.llvm.org/D158566 |
See also #59644, #50117, clangd/clangd#1712 |
Thanks for the report @hiraditya, the correctness issue is because we weren't asserting for executing all the work (hence we falsely, succeeded). Sent out https://reviews.llvm.org/D159337. Regarding the timeouts, the machine seems to be really busy again, building trivial preambles take ~11 secs: Considering these timeout issues are started happening recently, I wonder if we're spawning too many threads with the recent test execution strategy changes. Can you give some details of the machine you're running the tests/builds on @hiraditya ? |
We could fail going idle in 5 seconds and get spurious errors afterwards. See #64964. Differential Revision: https://reviews.llvm.org/D159337
There are some slow/congested bots, that can't go idle in 10 secs, see #64964 Differential Revision: https://reviews.llvm.org/D159338
I dont have the details, these are MacOS ci machines. Maybe there is something in the artifact directory you find helpful: https://ci.android.com/builds/submitted/10685355/darwin_mac/latest ? |
We could fail going idle in 5 seconds and get spurious errors afterwards. See llvm#64964. Differential Revision: https://reviews.llvm.org/D159337
There are some slow/congested bots, that can't go idle in 10 secs, see llvm#64964 Differential Revision: https://reviews.llvm.org/D159338
We started seeing a lot of timeouts that align with the change in lit to execute gtests in shards. The logic there assumes tests are single-threaded, which is the case for most of the LLVM, hence they pick #shards ~ #cores (by slightly overshooting). There are enough unittests in clangd that rely on multi-threading, they can create arbitrarily many threads but we limit amount of meaningful work to ~4 thread per process. This change ensures that we're accounting for that paralelism when executing clangd tests and not overloading test executors. In theory the change overestimates the requirements, not all tests are multi-threaded, but it doesn't seem to be resulting in any regressions on my local runs. Fixes llvm#64964. Fixes clangd/clangd#1712.
We could fail going idle in 5 seconds and get spurious errors afterwards. See llvm/llvm-project#64964. Differential Revision: https://reviews.llvm.org/D159337
There are some slow/congested bots, that can't go idle in 10 secs, see llvm/llvm-project#64964 Differential Revision: https://reviews.llvm.org/D159338
We started seeing a lot of timeouts that align with the change in lit to execute gtests in shards. The logic there assumes tests are single-threaded, which is the case for most of the LLVM, hence they pick #shards ~ #cores (by slightly overshooting). There are enough unittests in clangd that rely on multi-threading, they can create arbitrarily many threads but we limit amount of meaningful work to ~4 thread per process. This change ensures that we're accounting for that paralelism when executing clangd tests and not overloading test executors. In theory the change overestimates the requirements, not all tests are multi-threaded, but it doesn't seem to be resulting in any regressions on my local runs. Fixes llvm/llvm-project#64964. Fixes clangd/clangd#1712.
For more logs see for example:
Test logs pasted here.
In another run:
The text was updated successfully, but these errors were encountered: