Revert "Preventing containers from being unable to be deleted"#5153
Revert "Preventing containers from being unable to be deleted"#5153cyphar merged 1 commit intoopencontainers:mainfrom
Conversation
This fixes random failures to start a container in conmon integration tests (see issue 5151). I guess we need to find another way to fix issue 4645. This reverts commit 1b39997. Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
|
A quick (under a minute) repro, from runc top source dir: make
export RUNTIME_BINARY=$(pwd)/runc
$RUNTIME_BINARY --version
git clone https://github.com/containers/conmon && (cd conmon && make)
./conmon/test/run-tests.sh -j $(nproc)(comment out |
|
Or maybe conmon tests or conmon itself is doing something nasty; haven't figured it out yet. In any case it's a regression. |
rata
left a comment
There was a problem hiding this comment.
LGTM, thanks for investigating and finding a fix!
Is it always the same test failing? In that case, maybe we can adapt it and add it to runc?
|
@HirazawaUi you worked on the commit we are now reverting due to a regression. Wanna take a look on a way to fix it that doesn't regress? The repro for the regression is here: #5153 (comment) |
The failures are random, I haven't got to the bottom of it, but even if it is something not entirely kosher that conmon (or conmon tests) do (which results in a race), it worked before and it works with crun so this is a regression. As for adding runc tests, I was just thinking we can run existing conmon tests. The downside it when conmon tests will break, our CI will break, too, but I guess we can live with that (same as we're living with occasional CRIU failures). Similarly, we can add podman tests (crun does it already), but those are more heavy weight. |
|
This fixes random failures to start a container in conmon integration tests (see #5151).
Fixes: #5151.
I guess we need to find another way to fix #4645.
This reverts commit 1b39997 / PR #4757.