[LU-15022] sanity test_104c returned 2 Created: 21/Sep/21 Updated: 22/Sep/21 Resolved: 22/Sep/21 |
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Minor |
| Reporter: | Maloo | Assignee: | Arshad Hussain |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Issue Links: |
|
||||||||
| Severity: | 3 | ||||||||
| Rank (Obsolete): | 9223372036854775807 | ||||||||
| Description |
|
This issue was created by maloo for S Buisson <sbuisson@ddn.com> This issue relates to the following test suite run: https://testing.whamcloud.com/test_sets/4824ed24-8a35-468e-b1f1-3e1efcc5321f test_104c failed with the following error: test_104c returned 2 The test seems broken. After recordsize change lfs output : filesystem_summary: 73.0G 150.8M 72.9G 1% /mnt/lustre df output : 10.9.4.19@tcp:/lustre 74G 150M 73G 1% /mnt/lustre CMD: trevis-71vm4 /usr/sbin/lctl get_param -n osd-zfs.lustre-MDT0000..mntdev CMD: trevis-71vm4 zfs set recordsize=131072 lustre-mdt1/mdt1 CMD: trevis-71vm5 /usr/sbin/lctl get_param -n osd-zfs.lustre-MDT0000..mntdev trevis-71vm5: error: get_param: param_path 'osd-zfs/lustre-MDT0000//mntdev': No such file or directory pdsh@trevis-71vm1: trevis-71vm5: ssh exited with exit code 2 But when it succeeds, for instance in https://testing.whamcloud.com/sub_tests/272da434-33ef-44cc-9710-77a6372826e6, what happens is not really different, but the script manages to set recordsize: After recordsize change lfs output : filesystem_summary: 73.0G 146.8M 72.9G 1% /mnt/lustre df output : 10.9.10.22@tcp:/lustre 74G 147M 73G 1% /mnt/lustre CMD: trevis-209vm14 lctl get_param -n osd-zfs.lustre-MDT0000..mntdev CMD: trevis-209vm14 zfs set recordsize=131072 lustre-mdt1/mdt1 CMD: trevis-209vm15 lctl get_param -n osd-zfs.lustre-MDT0000..mntdev trevis-209vm15: error: get_param: param_path 'osd-zfs/lustre-MDT0000//mntdev': No such file or directory pdsh@trevis-209vm11: trevis-209vm15: ssh exited with exit code 2 CMD: trevis-209vm15 zfs set recordsize=131072 VVVVVVV DO NOT REMOVE LINES BELOW, Added by Maloo for auto-association VVVVVVV |
| Comments |
| Comment by Andreas Dilger [ 21/Sep/21 ] |
|
This started on 2021-09-14. The test was added in arch https://review.whamcloud.com/43154 " The error reported is because "ostparam" and "mdtparam" have a trailing "." but also adds its own "." when it is used. |
| Comment by Arshad Hussain [ 21/Sep/21 ] |
|
Andreas, I am looking into this. |
| Comment by Andreas Dilger [ 21/Sep/21 ] |
|
Actually, it may not be caused by the double ".." in the parameter name. Since these are converted to "//" in the pathname, and lookups will eat multiple consecutive "//" in the path, that may just be a cosmetic, but harmless issue. As Sebastien points out, in the case where the test passes, the "zfs set recordsize" command is run at the end, and that hides the error returned from "lctl get_param osd-zfs.*.mntdev". If the first error is fixed, then the test should pass consistently. It probably also makes sense to move the "restore recordsize" to stack_trap, so that it is always restored even if the test calls "error" earlier. Note, however, that the stack_trap commands are still run in the context of the test, so they cannot return an error or the test will be considered a failure. |
| Comment by Andreas Dilger [ 21/Sep/21 ] |
|
I just looked at the patch list from "git log --oneline --after 2021-09-16 master" and I think this is pretty clearly caused by patch https://review.whamcloud.com/44882 "LU-14997 tests: Register stack_trap for sanity/104c". Unfortunately, that patch was tested with "Test-Parameters: trivial" but test_104c is skipped for ldiskfs, and the patch causes the failure exclusively on ZFS backends. |
| Comment by Gerrit Updater [ 21/Sep/21 ] |
|
"Andreas Dilger <adilger@whamcloud.com>" uploaded a new patch: https://review.whamcloud.com/45008 |
| Comment by Gerrit Updater [ 22/Sep/21 ] |
|
"Oleg Drokin <green@whamcloud.com>" merged in patch https://review.whamcloud.com/45008/ |