Closed geissonator closed 5 years ago
Hung our CI last night for 10 hours :(
01:30:00 NOTE: Tasks Summary: Attempted 5391 tasks of which 4177 didn't need to be rerun and all succeeded.
01:30:00 + cp -r /tmp/openbmc/abi_version /tmp/openbmc/buildstats /tmp/openbmc/cache /tmp/openbmc/deploy /tmp/openbmc/hosttools /tmp/openbmc/log /tmp/openbmc/pkgdata /tmp/openbmc/saved_tmpdir /tmp/openbmc/sstate-control /tmp/openbmc/stamps /tmp/openbmc/sysroots /tmp/openbmc/sysroots-components /tmp/openbmc/work /tmp/openbmc/work-shared /var/lib/jenkins-slave/workspace/openbmc-build-gerrit-trigger-meta/distro/ubuntu/label/builder/target/zaius/openbmc/build/tmp
11:21:53 /var/lib/jenkins-slave/workspace/openbmc-build-gerrit-trigger-meta/distro/ubuntu/label/builder/target/zaius/openbmc-build-scripts/build-setup.sh: line 353: 26738 Terminated docker run --cap-add=sys_admin --net=host --rm=true -e WORKSPACE=${WORKSPACE} -w "${HOME}" -v "${HOME}":"${HOME}" ${mount_obmc_dir} ${mount_ssc_dir} --cpus="$num_cpu" -t ${img_name} ${WORKSPACE}/build.sh
11:21:53 Set build name.
11:21:53 New build name is '#125-Brad Bishop-openbmc/meta-inventec'
11:21:53 Build was aborted
11:21:53 Archiving artifacts
11:21:53 [WS-CLEANUP] Deleting project workspace...[WS-CLEANUP] done
11:21:53 Finished: ABORTED
A quick fix is to add a "timeout 300 cp ...." to enforce a max 5 minute timeout. We also should look into only copying off what the jenkins job needs by default (i.e. the deploy/images/** files.)
Those both sounds like pretty easy changes to do. I can work on that.
Charlie's fixes above have made this issue go away.
See this 1-2 times a day in our CI jobs. Then end of the build-setup.sh, where we're copying the files from the docker container to the mounted filesystem, it just hangs. You have to manually kill the job or it holds up all the other jobs.
Here's an example:
Seems like we could def optimize this to only copy of the artifacts we actually need. Maybe provide an option to copy it all, but default to just the image artifacts.