[LU-6275] wrong cleanup with LWP start. Created: 24/Feb/15  Updated: 16/May/16  Resolved: 10/Jul/15

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.5.0, Lustre 2.6.0
Fix Version/s: Lustre 2.8.0

Type: Bug Priority: Critical
Reporter: Alexey Lyashkov Assignee: WC Triage
Resolution: Fixed Votes: 0
Labels: patch

Severity: 3
Rank (Obsolete): 17595

 Description   

if any issues hit with processing a config to start LWP on MDT node, lsi_lwp_started don't set, so lustre_end_log will don't called.
it's wrong in rare case on current MGC code, and completely wrong for an new MGC cleanup.
for older code we may take an MGC lock, but failed with take a config from MGS. mgc will cancel a lock in that situation and add 'CLD'/lock to re-enqueue.
but shutdown (cleanup) started as lustre_start_lwp failed.
cleanup forget to call lustre_end_log as lsi_lwp_started don't set - BUT lock/cld still on enqueue list.. OOPS. next re-enqueue loop will found a cld and start enqueue and will failed on dereference for superblock as cld_stopping isn't set.

bug looks exist for any code, but more easily hit from xyratex mgc cleanup.



 Comments   
Comment by Gerrit Updater [ 25/Feb/15 ]

Alexey Lyashkov (alexey.lyashkov@seagate.com) uploaded a new patch: http://review.whamcloud.com/13863
Subject: LU-6275 obdclass: don't forget to remove LWP mgc config
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 9c88e5d689281a7519c5facd784e08812075cdd7

Comment by Gerrit Updater [ 10/Jul/15 ]

Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/13863/
Subject: LU-6275 obdclass: don't forget to remove LWP mgc config
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 42b7ddf41614c4cdf9b7d54758636c319d1560da

Comment by Peter Jones [ 10/Jul/15 ]

Landed for 2.8

Generated at Sat Feb 10 01:58:49 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.