[LU-3715] test-framework making ldiskfs specific calls without checking for zfs Created: 06/Aug/13 Updated: 11/Sep/13 Resolved: 11/Sep/13 |
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | Lustre 2.5.0 |
| Fix Version/s: | Lustre 2.5.0 |
| Type: | Bug | Priority: | Minor |
| Reporter: | Keith Mannthey (Inactive) | Assignee: | Bruno Faccini (Inactive) |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | zfs | ||
| Severity: | 3 |
| Rank (Obsolete): | 9568 |
| Description |
|
As seen from a review-zfs run This is the tail end of check_and_setup_lustre 23:13:45:CMD: wtm-10vm2.rosso.whamcloud.com lctl dl | grep ' IN osc ' 2>/dev/null | wc -l
23:13:45:disable quota as required
23:13:45:CMD: wtm-10vm7,wtm-10vm8 /usr/sbin/lctl set_param osd-ldiskfs.track_declares_assert=1 || true
23:13:45:wtm-10vm7: error: set_param: /proc/{fs,sys}/{lnet,lustre}/osd-ldiskfs/track_declares_assert: Found no match
23:13:45:wtm-10vm8: error: set_param: /proc/{fs,sys}/{lnet,lustre}/osd-ldiskfs/track_declares_assert: Found no match
23:13:45:
23:13:46:== runtests test 1: All Runtests == 23:13:45 (1375337625)
23:13:46:touching /mnt/lustre at Wed Jul 31 23:13:45 PDT 2013
from check_and_setup_lustre if [ $(lower $OSD_TRACK_DECLARES_LBUG) == 'yes' ] ; then
local facets="$(get_facets OST),$(get_facets MDS),mgs"
local nodes="$(facets_hosts ${facets})"
if [ -n "$nodes" ] ; then
do_nodes $nodes "$LCTL set_param \
osd-ldiskfs.track_declares_assert=1 || true"
fi
fi
osd-ldiskfs only works on ldiskfs based systems. |
| Comments |
| Comment by Keith Mannthey (Inactive) [ 06/Aug/13 ] |
|
Bruno it looks like this is your code. |
| Comment by Bob Glossman (Inactive) [ 06/Aug/13 ] |
|
I think this change from Bruno was deliberate. commit b07f02551285b1f1ad8c193cf5c7c5b10e33ac39 from Jun 18 removed the check for ldiskfs only, but also arranged for the error return from the set_param osd-ldiskfs.track_declares_assert=1 to be ignored. I agree it would have been nicer and cleaner to do the set_param only on ldiskfs nodes, but the way it is looks to have been done on purpose. |
| Comment by Keith Mannthey (Inactive) [ 06/Aug/13 ] |
|
I remember this now a little. |
| Comment by Andreas Dilger [ 08/Aug/13 ] |
|
It is my fault that this error is being seen. It is intentionally harmless, though it would be possible to send the error to /dev/null. |
| Comment by Keith Mannthey (Inactive) [ 08/Aug/13 ] |
|
What if we we are on ldiskfs and the things really fails? Could we leave a comment in the test so it clear this is really what we want to do? |
| Comment by Andreas Dilger [ 09/Aug/13 ] |
|
This setting is just for debugging that we might need in very rare cases. If it actually fails, it won't hurt anything, and we probably won't care very much. |
| Comment by Andreas Dilger [ 20/Aug/13 ] |
| Comment by Bruno Faccini (Inactive) [ 20/Aug/13 ] |
|
Thanks to make it Andreas! |
| Comment by Andreas Dilger [ 11/Sep/13 ] |
|
Patch landed for 2.4.93. |