[LU-9406] conf-sanity test 82b fails with ‘FAIL: /usr/sbin/lctl pool_add scratch.test_82b scratch-OST[*] failed’ Created: 26/Apr/17  Updated: 17/Jun/19  Resolved: 17/Jun/19

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.10.0
Fix Version/s: Lustre 2.11.0

Type: Bug Priority: Minor
Reporter: James Nunez (Inactive) Assignee: James Nunez (Inactive)
Resolution: Fixed Votes: 0
Labels: test
Environment:

A Lustre configuration with the MDS and the MGS on separate nodes


Issue Links:
Related
is related to LU-8688 All Lustre test suites should run/PAS... Open
Severity: 3
Rank (Obsolete): 9223372036854775807

 Description   

conf-sanity test_82b fails when Lustre is configured with the MGS and MDS on separate nodes with

conf-sanity test_82b: @@@@@@ FAIL: /usr/sbin/lctl pool_add scratch.test_82b scratch-OST[32b4,06a8,7070] failed

From the console log, it looks like the OST pools are never created

eagle-48vm1: Warning, pool scratch.test_82b not found
eagle-48vm1: Pool scratch.test_82b not found
eagle-48vm1: Pool scratch.test_82b not found
eagle-48vm1: Pool scratch.test_82b not found
eagle-48vm1: pool_add: No such file or directory
pdsh@eagle-48vm6: eagle-48vm1: ssh exited with exit code 2
 conf-sanity test_82b: @@@@@@ FAIL: /usr/sbin/lctl pool_add scratch.test_82b scratch-OST[32b4,06a8,7070] failed 

test 82b checks that you can create OST pools with the –ost-list option. For OST pools to work with the MDS and the MGS on separate nodes, there must be a client running on the MGS. The problem is that all clients and servers are stopped at the beginning of the test and then only a single client is started. That single client may or may not be the client on the MGS.

From conf-sanity test 82b:

5846         # Setup Lustre filesystem.
5847         start_mgsmds || error "start_mgsmds failed"
5848         for i in $(seq $OSTCOUNT); do
5849                 start ost$i $(ostdevname $i) $OST_MOUNT_OPTS ||
5850                         error "start ost$i failed"
5851         done
5852 
5853         mount_client $MOUNT || error "mount client $MOUNT failed"

One solution to this problem is to restart all clients in this test.



 Comments   
Comment by Gerrit Updater [ 26/Apr/17 ]

James Nunez (james.a.nunez@intel.com) uploaded a new patch: https://review.whamcloud.com/26844
Subject: LU-9406 test: conf-sanity 82b start all clients
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: a353f402a69e9953c32bda118da43cbdd835c8d6

Comment by James Nunez (Inactive) [ 18/Oct/17 ]

conf-sanity test 82 is fixed in patch https://review.whamcloud.com/#/c/28806

Patch https://review.whamcloud.com/26844 was abandoned.

Comment by James Nunez (Inactive) [ 17/Jun/19 ]

conf-sanity test 82b was fixed by restarting all clients when testing with a separate MGS.

Generated at Sat Feb 10 02:25:55 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.