NFS Restart Needed after Replication

[Hellrazorx]

Hi,

Heres is the setup:

• Two Pools on the same Truenas system A and B
• One dataset is replicated from A to B with a daily replication task
• Dataset on Pool B has a functional nfs share (read only). Shared to a client

Everytime replication task runs. Nfs share access to the client is lost.

Disabling the NFS service and turning it back on make it work again.

Is this behavior known? Is there a parameter that I'm not aware of?

Or is there a possible scripted workaround?

Thanks

 

[Kris Moore]

That is a weird failure... We'd probably want to look at a bug fix for that. Can you please file a ticket on https://jira.ixsystems.com and include a debug file from the system? (System -> Advanced -> Save Debug)

 

[Hellrazorx]

I made a Cron Job for now
/etc/rc.d/ nfsd restart
Scheduled half an hour after the replication process.
Looked today and the NFS link seemed healthy

I'll fill your ticket anyway

EDIT: Nope, it doesn't work.

 

[kajer]

It's not NFS, it's mountd

if you restart mountd post replication, the shares will work again.

I have had to change my twice daily replication to once at 5am and mountd restart by 7am so I can use KODI via NFS against the read only dataset I'm replicating from another array. Previously, I had the source Truenas box PUSH replicating to the target box and never had NFS issues at all. I recently re-installed 12.8u1 on the source box, and converted the local instances to PULL from said source.

Whatever is happening on the replication task, the mountpoint for the NFS share changes underneath the NFS / mountd service and breaks the share until mountd is restarted.