SOLVED NFS service not starting at boot missing /etc/exports

[aacebedo]

Hi all

Just migrated a VM from CORE to SCALE by rebuilding it.
Everything worked nice except the NFS service which is not starting at first boot.

It complains about not finding /etc/exports file (despite the fact I have configured some shares in the GUI) which seems to be generated by the middlewared service. I guess the nfs server service is started before the middlewared one. If I restart the NFS service manually through SSH it works (the /etc/exports file is then present).
Here is the log of the nfs server:

Code:

● nfs-server.service - NFS server and services
     Loaded: loaded (/lib/systemd/system/nfs-server.service; enabled; vendor preset: disabled)
     Active: failed (Result: exit-code) since Fri 2022-03-04 23:37:38 CET; 14s ago

Mar 04 23:37:38 <REDACTED> systemd[1]: Starting NFS server and services...
Mar 04 23:37:38 <REDACTED> exportfs[3754]: exportfs: can't open /etc/exports for reading
Mar 04 23:37:38 <REDACTED> systemd[1]: nfs-server.service: Control process exited, code=exited, status=1/FAILURE
Mar 04 23:37:38 <REDACTED> systemd[1]: nfs-server.service: Failed with result 'exit-code'.
Mar 04 23:37:38 <REDACTED> systemd[1]: Stopped NFS server and services.

 

[Kris Moore]

Did you skim through your specific share configurations? Ref: https://www.truenas.com/docs/releasenotes/scale/22.02.0/#known-issues

 

[aacebedo]

Thank you for the answer.
I have recreated completely the install (did not migrated any conf from CORE)
When I restart manually the service without changing the configuration, I am able to mount the export that's not an issue.
The issue is the /etc/exports file does not seem do be available when the machine boots and become available a little bit after.
I also tried to remove and readd the share in config without success.

 

[aacebedo]

I think I found something, my ZFS pool is plugged onto a SCSI card in passtrough and this log makes me thing maybe the pool has not been mounted yet when the middlewared is started:

Code:

 - /mnt/<POOL_NAME>/<DATASET>: path does not exist, omitting from NFS exports

When the machine has finished its booting process, the datapool is mounted and accessible.

 

[anodos]

I think I found something, my ZFS pool is plugged onto a SCSI card in passtrough and this log makes me thing maybe the pool has not been mounted yet when the middlewared is started:

Code:

 - /mnt/<POOL_NAME>/<DATASET>: path does not exist, omitting from NFS exports

When the machine has finished its booting process, the datapool is mounted and accessible.

Okay. You can probably add a post-init script to `midclt call service.restart nfs`

 

[aacebedo]

Okay. You can probably add a post-init script to `midclt call service.restart nfs`

That's working ! Thanks
However don't you think the start of the nfs service shall wait for the zpool to be available ? I guess it can happen sometimes especially if it depends on the speed of the machine executing truenas

 

[anodos]

That's working ! Thanks
However don't you think the start of the nfs service shall wait for the zpool to be available ? I guess it can happen sometimes especially if it depends on the speed of the machine executing truenas

Right, it's a race condition that we'll need to fix. I proposed a workaround for you for now :)

 

[aacebedo]

Right, it's a race condition that we'll need to fix. I proposed a workaround for you for now :)

Thank you :) Shall I create a ticket in the TrueNAS bug report ?

 

[soleous]

Not to add any valuable contribution, but I also experienced this. It only happened once, additional reboots have not resulted in NFS failing to start.

It is also in a VM scenario with pass though.