CI: make systemd tests parallel-safe (*) #24048
Conversation
openshift-ci
bot
added
release-note-none
approved
|
LGTM |
test/system/250-systemd.bats
Outdated
| mv -Z "container-$cname.service" $UNIT_FILE.tmp.$$ && \ | ||
| mv -Z $UNIT_FILE.tmp.$$ $UNIT_FILE |
There was a problem hiding this comment.
I am confused, why do we move to a tmp file just to move to the real location directly after? I do not see th epoint of doing this vs one single mv command?
There was a problem hiding this comment.
Ultraparanoia. mv is not atomic across filesystems, so this is theoretically possible:
- mv starts writing foo.service
- another parallel job runs systemctl daemon-reload
- systemd sees (incomplete) foo.service and does whatever systemd does with it and bad things happen
- mv completes
Even with all this paranoia, #24010 still triggers (on my laptop), so I'm pretty sure that the fragment-file flake has nothing to do with corrupt systemd service files ... but I just want to eliminate every last possibility of test errors.
There was a problem hiding this comment.
Ok got it but then please comment the reason here because it is not obvious otherwise.
test/system/250-systemd.bats
Outdated
| mv -Z "container-$cname.service" $TEMPLATE_FILE.tmp.$$ && \ | ||
| mv -Z $TEMPLATE_FILE.tmp.$$ $TEMPLATE_FILE |
Mostly just switch to safename. Rewrite setup() to guarantee unique service file names, atomically created. * IMPORTANT NOTE: enabling parallelization on these tests triggers containers#24010 ("fragment file" flake), but only on my f40 laptop. I have never seen the flake in Cirrus despite many many runs in containers#23275. I am submitting this for review and merging because even though _something_ is broken, this breakage is unlikely to affect our CI. Signed-off-by: Ed Santiago <[email protected]>
openshift-ci
bot
added
the
do-not-merge/work-in-progress
edsantiago
force-pushed
the
safename-250
branch
from
01f5053
to
faf4604
Compare
openshift-ci
bot
removed
the
do-not-merge/work-in-progress
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: edsantiago, Luap99 The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
openshift-merge-bot
bot
merged commit f9f72f5
into
containers:main
Mostly just switch to safename. Rewrite setup() to guarantee
unique service file names, atomically created.
triggers unable to create pod cgroup: slice was already loaded or has a fragment file #24010 ("fragment file" flake), but only on my
f40 laptop. I have never seen the flake in Cirrus despite
many many runs in WIP: system test parallelization: two-pass approach #23275. I am submitting this for review
and merging because even though something is broken,
this breakage is unlikely to affect our CI.
Signed-off-by: Ed Santiago [email protected]