[LU-4879] sanity-lfsck test 17 failure: test_17 failed with 3

Related — Mon, 7 Jul 2014 21:46:51 +0000

Sanity-lfsck test 17 has failed with error message "test_17 failed with 3" a few times over the past month. Results from one failure are at https://maloo.whamcloud.com/sub_tests/2c265c8e-c040-11e3-8c28-52540035b04c .

The test 17 creates an MDT-OST inconsistency using fail_loc=0x1614 (OBD_FAIL_LFSCK_MULTIPLE_REF) meaning two MDT objects reference the same OST object. LFSCK layout is called and is waiting for a status of "complete", but LFSCK finished with a status of "partial". From the client test_log:

Started LFSCK on the device lustre-MDT0000: layout.
CMD: client-30vm3 /usr/sbin/lctl get_param -n 		mdd.lustre-MDT0000.lfsck_layout |
		awk '/^status/ { print \$2 }'
CMD: client-30vm3 /usr/sbin/lctl get_param -n 		mdd.lustre-MDT0000.lfsck_layout |
		awk '/^status/ { print \$2 }'
Waiting 3 secs for update
CMD: client-30vm3 /usr/sbin/lctl get_param -n 		mdd.lustre-MDT0000.lfsck_layout |
		awk '/^status/ { print \$2 }'
CMD: client-30vm3 /usr/sbin/lctl get_param -n 		mdd.lustre-MDT0000.lfsck_layout |
		awk '/^status/ { print \$2 }'
CMD: client-30vm3 /usr/sbin/lctl get_param -n 		mdd.lustre-MDT0000.lfsck_layout |
		awk '/^status/ { print \$2 }'
Update not seen after 3s: wanted 'completed' got 'partial'
 sanity-lfsck test_17: @@@@@@ FAIL: test_17 failed with 3

On the MDS console, after LFSCK is started, we see:

13:27:59:LustreError: 30778:0:(lfsck_layout.c:1422:lfsck_layout_master_notify_others()) lustre-MDT0000-osd: fail to notify OST 7 for layout start: rc = -95

Whamcloud Community JIRA

[LU-4879] sanity-lfsck test 17 failure: test_17 failed with 3