[LU-15145] hsm_cancel on an inactive HSM restore request do not free the EX lock Created: 22/Oct/21  Updated: 26/Oct/22  Resolved: 18/Jan/22

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: None
Fix Version/s: Lustre 2.15.0

Type: Bug Priority: Major
Reporter: Etienne Aujames Assignee: Etienne Aujames
Resolution: Fixed Votes: 0
Labels: None
Environment:

VMs + Lustre 2.14.55_43_g6a08df2
lhsmtool_posix

Issue Links:
Related
is related to LU-15132 Parallel data accesses on a release f... Resolved
Severity: 3
Rank (Obsolete): 9223372036854775807

 Description   

For an inactive HSM RESTORE request (not sent to a copytool), CANCEL action (or purge) only delete the record inside the llog but it does not unlock the MDS_INODELOCK_LAYOUT exclusive lock.

This causes an orphan EX lock on the fid.
Combined with the LU-15132, it could hang easily entirely a MDT.

 Comments   
Comment by Gerrit Updater [ 22/Oct/21 ]

"Etienne AUJAMES <eaujames@ddn.com>" uploaded a new patch: https://review.whamcloud.com/45341
Subject: LU-15145 hsm: unlock the restore layout lock for a cancel
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 820f0d18bc23cfbb095249dad328b3665d6988ad

Comment by Gerrit Updater [ 18/Jan/22 ]

"Oleg Drokin <green@whamcloud.com>" merged in patch https://review.whamcloud.com/45341/
Subject: LU-15145 hsm: unlock the restore layout lock for a cancel
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 6d4019281b392bcb6993d1cfca3d47d7fa5f7c56

Comment by Gerrit Updater [ 18/Jan/22 ]

"Etienne AUJAMES <eaujames@ddn.com>" uploaded a new patch: https://review.whamcloud.com/46168
Subject: LU-15145 hsm: unlock the restore layout lock for a cancel
Project: fs/lustre-release
Branch: b2_12
Current Patch Set: 1
Commit: 0ab58a57c1dc3d71d44885c07cad81ec142d2694

Comment by Peter Jones [ 18/Jan/22 ]

Landed for 2.15

Generated at Sat Feb 10 03:15:50 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.