Tests for disconnections during lxc exec
#284
Conversation
|
There are too many commits in here, @hamistao did something went wrong when pushing/opening the PR? |
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
from
f5ae15b to
7a05cb3
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
2 times, most recently
from
9a612c8 to
53d9683
Compare
|
@simondeziel Yes I messed up my rebase, I should have opened this as a draft. Sorry about that. |
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
5 times, most recently
from
0a38f38 to
ec1e9dc
Compare
|
@simondeziel Fixed, this should not be merged yet since the VM fix tested here is not in LXD yet, but feel free to take a look whenever you can. |
No worries at all! |
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
from
ec1e9dc to
2c8a847
Compare
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
from
2c8a847 to
c96f034
Compare
|
@simondeziel happy with this? |
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
2 times, most recently
from
42c07fc to
0ba22da
Compare
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
from
0ba22da to
2146528
Compare
|
@simondeziel This is ready for another review. I believe the failing tests are unrelated to these changes since all modified files' tests pass. |
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
3 times, most recently
from
2f00d90 to
707e85c
Compare
|
@simondeziel I keep getting a |
There was a problem hiding this comment.
https://github.com/canonical/lxd-ci/actions/runs/10885376800/job/30202865680?pr=284 taking ~2h feels like a bug as I'd expect lxc stop to timeout before that long, no?
That said, I think the context deadline bit might be addressed by waiting for complete boot before doing the shutdown.
tests/vm
Outdated
|
|
||
| # Launch test instance | ||
| lxc launch "${IMAGE}" vm1 --vm | ||
| waitInstanceReady vm1 |
There was a problem hiding this comment.
| waitInstanceReady vm1 | |
| waitInstanceBooted vm1 |
There was a problem hiding this comment.
taking ~2h feels like a bug as I'd expect lxc stop to timeout before that long, no?
I think the timeout is indeed very long so this should be expected
That said, I think the context deadline bit might be addressed by waiting for complete boot before doing the shutdown.
Yeah, that makes sense, thanks!
tests/vm
Outdated
| echo "==> Test cleanly stopping a VM" | ||
| lxc stop vm1 | ||
| lxc start vm1 | ||
| waitInstanceReady vm1 |
There was a problem hiding this comment.
| waitInstanceReady vm1 | |
| waitInstanceBooted vm1 |
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
from
707e85c to
b196f12
Compare
|
@simondeziel This is ready again |
Signed-off-by: hamistao <[email protected]>
Signed-off-by: hamistao <[email protected]>
Signed-off-by: hamistao <[email protected]>
Signed-off-by: hamistao <[email protected]>
Signed-off-by: hamistao <[email protected]>
Signed-off-by: hamistao <[email protected]>
hamistao
force-pushed
the
reboot_experience_improv_tests
branch
from
b196f12 to
8e4438c
Compare
As was discussed with @mihalicyn If containers are stopped gracefully, the exit code can be 129(SIGHUP) or 143(SIGTERM) depending on the signal used, both results have been seen and make sense. If forcefully stopping a container the exit code is always 137 (SIGKILL). For VMs, the exit code was standardized as 129 for all cases by #13875 (not merged yet).
In its current state, this PR only adds tests for forcefully stopping containers and VMs.