[LU-5830] use lustre_rmmod in lnet start/stop script Created: 31/Oct/14  Updated: 06/May/17  Resolved: 17/Nov/14

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.7.0, Lustre 2.5.4
Fix Version/s: Lustre 2.7.0, Lustre 2.5.4

Type: Bug Priority: Minor
Reporter: Bruno Faccini (Inactive) Assignee: Bruno Faccini (Inactive)
Resolution: Fixed Votes: 0
Labels: None

Issue Links:
Blocker
Duplicate
duplicates LU-2669 Init script "lnet stop" fails due to ... Resolved
Related
Severity: 3
Rank (Obsolete): 16352

 Description   

When trying to enable and use lnet start/stop script to shutdown a stand-alone Lustre Server it fails with errors during modules unload. This comes from a wrong list/order of modules to be unloaded. It would be safer to use lustre_rmmod to do so instead.
Master patch is at http://review.whamcloud.com/12513.

 Comments   
Comment by Jian Yu [ 05/Nov/14 ]

Here is the back-ported patch for Lustre b2_5 branch: http://review.whamcloud.com/12587

Comment by Peter Jones [ 17/Nov/14 ]

Landed for 2.7

Comment by Andreas Dilger [ 01/Dec/14 ]

Bruno, I noticed that this patch was very similar to http://review.whamcloud.com/5478 "LU-2669 scripts: Change behavior of lustre_rmmod" but doesn't do quite the same thing. That patch removed the Lustre modules before stopping LNet, then removed the ptlrpc/lnet/libcfs modules afterward.

Is there still value in the changes to lustre_rmmod that http://review.whamcloud.com/5478 made, or should it be abandoned?

Comment by Bruno Faccini (Inactive) [ 01/Dec/14 ]

Andreas, you are right they are very similar, and particularly both tickets reason since they both bring out the same flaw in lnet shutdown script. The main difference comes from the need to strictly follow the 3 steps procedure that you have described, which I have never found to be necessary, but may be I am wrong?
Also last patch-set of change #5478 for LU-2669 has triggered a lot of tests failures that may be caused by the changes it applies to lustre_rmmod, when my patch has not encountered failures during auto-tests.

Comment by Andreas Dilger [ 02/Dec/14 ]

Bruno, have you ever done testing with LNet routing to see if that could cause problems that aren't seen in our regular testing? I can't imagine that this three-step shutdown process was created unless there was a reason to do so.

Comment by Gerrit Updater [ 15/Jan/15 ]

Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/12587/
Subject: LU-5830 scripts: use lustre_rmmod in lnet start/stop script
Project: fs/lustre-release
Branch: b2_5
Current Patch Set:
Commit: e5966bdbb7c7644cc04580ad6b4a9006def3f758

Generated at Sat Feb 10 01:54:53 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.