Bug #44790
closedqa: adjust-ulimits invoked after being removed
0%
Description
2020-03-26T16:57:22.660 DEBUG:teuthology.run_tasks:Unwinding manager install 2020-03-26T16:57:22.677 INFO:teuthology.task.install.util:Removing shipped files: /home/ubuntu/cephtest/valgrind.supp /usr/bin/daemon-helper /usr/bin/adjust-ulimits... 2020-03-26T16:57:22.677 INFO:teuthology.orchestra.run.smithi029:> sudo rm -f -- /home/ubuntu/cephtest/valgrind.supp /usr/bin/daemon-helper /usr/bin/adjust-ulimits 2020-03-26T16:57:22.681 INFO:teuthology.orchestra.run.smithi136:> sudo rm -f -- /home/ubuntu/cephtest/valgrind.supp /usr/bin/daemon-helper /usr/bin/adjust-ulimits ... 2020-03-26T16:59:29.422 DEBUG:teuthology.run_tasks:Unwinding manager ceph 2020-03-26T16:59:29.447 INFO:tasks.ceph.ceph_manager.ceph:Canceling any pending splits or merges... 2020-03-26T16:59:29.447 INFO:teuthology.orchestra.run.smithi136:> sudo adjust-ulimits ceph-coverage /home/ubuntu/cephtest/archive/coverage timeout 120 ceph --cluster ceph --log-early osd dump --format=json 2020-03-26T16:59:29.473 INFO:teuthology.orchestra.run.smithi136.stderr:sudo: adjust-ulimits: command not found
Updated by Greg Farnum about 4 years ago
This popped up a couple times many years ago; I'm not sure if it's a very rare race in shutdown or what. No apparent changes in the FS qa suites that could have caused it.
Updated by Greg Farnum about 4 years ago
Happened again in /a/gregf-2020-04-01_18:45:43-fs-wip-greg-testing-41-distro-basic-smithi/4914097
I did happen to notice an error in the log right before it started unwinding, but it looks like the test was already in shutdown mode and I'm not sure if it's relevant:
2020-04-01T20:45:17.443 WARNING:tasks.check_counter:Counter 'mds.dir_split' not found on daemon mds.b
2020-04-01T20:45:17.443 DEBUG:teuthology.run_tasks:Unwinding manager install
2020-04-01T20:45:17.461 INFO:teuthology.task.install.util:Removing shipped files: /home/ubuntu/cephtest/valgrind.supp /usr/bin/daemon-helper /usr/bin/adjust-ulimits...
Updated by Greg Farnum about 4 years ago
- Priority changed from Normal to High
Updated by Greg Farnum about 4 years ago
http://pulpito.ceph.com/yuriw-2020-04-07_17:47:23-fs-wip-octopus-rgw-msg-fixes-distro-basic-smithi/4931869/
http://pulpito.ceph.com/yuriw-2020-04-07_17:47:23-fs-wip-octopus-rgw-msg-fixes-distro-basic-smithi/4931949/
These are both fsx runs and it wasn't just adjust-ulimits, hit it with fusermount as well. But still seems to be we are somehow doing teardown out of order.
Updated by Patrick Donnelly about 4 years ago
- Is duplicate of Bug #45387: qa: install task runs twice with double unwind causing fatal errors added