alamb commented on pull request #9653:
URL: https://github.com/apache/arrow/pull/9653#issuecomment-792320959
> We've already observed from our local repro attempts that there wasn't any
specific PR that caused the issue. It could be that some dependency updates
suddenly caused us to hit the memory limits on CI, as you're exploring.
I think this is a likely theory but I don't have any specific data to prove
one way or the other
> Curious, when the CI fails, we will no longer have the trace, right? Isn't
that a concern whenever we want to share logs e.g. on PRs, mailing lists, etc?
I think we will still get backtraces even without "debug" symbols (the debug
symbols are used by gdb / lldb to be able to interpret core dumps / map local
variables to memory locations, etc).
To confirm this theory I ran the following (on my mac, will try on linux
shortly) and the trace is still present
```
(arrow_dev) alamb@MacBook-Pro:~/Software/arrow2/rust$ RUST_BACKTRACE=1
RUSTFLAGS="-C debuginfo=0" cargo test -p arrow -- test_null_list_primitive
Finished test [unoptimized + debuginfo] target(s) in 0.10s
Running target/debug/deps/arrow-99caf14857691f1e
running 1 test
test array::array::tests::test_null_list_primitive ... FAILED
failures:
---- array::array::tests::test_null_list_primitive stdout ----
thread 'array::array::tests::test_null_list_primitive' panicked at 'test
error', arrow/src/array/array.rs:621:9
stack backtrace:
0: std::panicking::begin_panic
1: arrow::array::array::tests::test_null_list_primitive
2: arrow::array::array::tests::test_null_list_primitive::{{closure}}
3: core::ops::function::FnOnce::call_once
4: core::ops::function::FnOnce::call_once
at
/rustc/cb75ad5db02783e8b0222fee363c5f63f7e2cf5b/library/core/src/ops/function.rs:227:5
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose
backtrace.
failures:
array::array::tests::test_null_list_primitive
test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 667 filtered
out; finished in 0.14s
error: test failed, to rerun pass '-p arrow --lib'
```
> A compromise could be to disable debug symbols only on the Debian Rust
stable runs, as the other platforms seem fine.
Yeah -- in fact this PR doesn't set the debug flags for Windows or Mac (I
believe the Mac builders are larger -- 15GB of memory on github, and windows
debugging symbols work significantly different)
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]