Bump similar_asserts to 1.6 #4269
Conversation
This allows us to remove a call to `String::leak` as they relaxed the life time requirements on the one method we use.
weiznich
force-pushed
the
bump/similar_assert
branch
from
4547bde to
325ffc5
Compare
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
weiznich
force-pushed
the
bump/similar_assert
branch
8 times, most recently
from
591dc12 to
52dc16f
Compare
weiznich
force-pushed
the
bump/similar_assert
branch
from
52dc16f to
33ff9ee
Compare
| @@ -636,6 +637,9 @@ mod tests { | |||
| assert_eq!(release_count.load(Ordering::Relaxed), 2); | |||
| assert_eq!(checkin_count.load(Ordering::Relaxed), 3); | |||
| assert_eq!(checkout_count.load(Ordering::Relaxed), 3); | |||
| // this is required to workaround a segfault while shutting down | |||
| // the pool | |||
| std::thread::sleep(Duration::from_millis(100)); | |||
There was a problem hiding this comment.
This is unfortunate. Do we know what we're waiting for? A timed delay like this has the potential to be unreliable, e.g., if the test system is slow for some reason. If possible it would be best to wait in a loop, testing for whatever condition is needed to avoid the segfault.
There was a problem hiding this comment.
You are correct that this is not great. Unfortunately I do not know what exactly we are waiting for, just that waiting seems to resolve the segfaults. My guess is that there is something happening in parallel here that causes the segfault. R2d2 is acquiring a new connection, while the pool is shutting down it might just be that openssl cannot handle creating/dropping a connection in a fast turnaround? Or that it returns early but does stuff on it's own in the background, which in turn leads to collisions while shutting down the connection?
Given that I did not found a way to reproduce the problem locally it's relatively hard for me to debug it. I just went for the first thing that seem to resolve the issue to unblock the other PR's for now. If someone has a better solution I'm more than happy to accept that.
This allows us to remove a call to
String::leakas they relaxed the life time requirements on the one method we use.