[ClusterLabs] Antw: [EXT] nfs mount resource won't stop - lazy option ?

Ulrich Windl Ulrich.Windl at rz.uni-regensburg.de
Mon Jul 11 08:53:42 EDT 2022


>>> Patrick Vranckx <patrick.vranckx at uclouvain.be> schrieb am 08.07.2022 um 11:25
in Nachricht <69422771-c57d-d910-8e92-8a2345847967 at uclouvain.be>:
> Hi,
> 
> I have a cluster configures as such:
> 
> Online: [ mbacktm1 mbacktm2 ]
> 
> Full list of resources:
> 
>   stonith-mbt    (stonith:fence_scsi):    Started mbacktm2
>   Resource Group: group-mbt-stor
>       nfs-daemon    (ocf::heartbeat:nfsserver):    Started mbacktm2
>       mount_admnfs    (ocf::heartbeat:Filesystem):    Started mbacktm2
>   Resource Group: group-secondary-gateway
>       mount_admnfs_1    (ocf::heartbeat:Filesystem):    Started mbacktm1
> 
> 
> A daemon NFS is started on one node, a filesystem is exported and 
> mounted on both node.
> 
> If the node running the nfs server crashes, reboots or is fenced, the 
> resource "mount_admnfs_1" on the other node won't stop because the nfs 
> server will not respond (unable to unmount the filesystem anymore)
> 
> In such a situation without pacemaker, one can do a "umount -l" to solve 
> the problem.
> 
> Is there a solution  for Pacemaker ?

What is your specific problem? NFS design is that clients wait for the server to re-appear.

> 
> Best regards,
> 
> Patrick
> 
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users 
> 
> ClusterLabs home: https://www.clusterlabs.org/ 






More information about the Users mailing list