[LU-14074] LNet: lustre_rmmod fails after lnetctl is used

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.15.0
• Severity: 3
• Assignee: Cyril Bordage
• Reporter: Amir Shehata (Inactive)

Description

After using the lnetctl utility a reference count is taken on the lnet modules and lnetctl lnet unconfigure needs to be called in order for lustre_rmmod to remove the lnet module.

I think it would be a good idea to modify lustre_rmmod to unconfigure lnet to avoid having to do it manually.

Steps to reproduce:

modprobe lustre
lnetctl lnet configure
lustre_rmmod 

Issue Links

• is related to: LU-14627 Lost ref on lnet_peer in discovery leads to LNetError: 24909:0:(peer.c:292:lnet_destroy_peer_locked()) ASSERTION( list_empty(&lp->lp_peer_nets) ) failed:

Comments

[Peter Jones]

Landed for 2.15

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/41698/
Subject: LU-14074 scripts: automatic LNet unconfigure
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: ebdf71a2df044c441ff633dbfc2f463520892bfb

[Gerrit Updater]

Cyril Bordage ([e-mail]) uploaded a new patch: https://review.whamcloud.com/41698
Subject: LU-14074 scripts: automatic LNet unconfigure
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 4a61ed82c999a8d4449407de19aa6524a20214cf

[Andreas Dilger]

I agree that this would be very useful. When this change is made, the code in lustre/scripts/lnet.in should be modified correspondingly:

-        lustre_rmmod ptlrpc || exit 1
-        lnetctl lnet unconfigure || exit 1
-        lustre_rmmod libcfs ldiskfs || exit 1
+       lustre_rmmod || exit 1

along with lustre/scripts/systemd/lnet.service.in:

-ExecStop=@sbindir@/lustre_rmmod ptlrpc
-ExecStop=@sbindir@/lnetctl lnet unconfigure
-ExecStop=@sbindir@/lustre_rmmod libcfs ldiskfs
+ExecStop=@sbindir@/lustre_rmmod