Autofs monitoring problems with NFS mount points on AIX partitions

[dposaric]

Hi guys,

I have a question about file system discovery and monitoring of autofs mount points on AIX partitions.

We have a large environment with about 400 AIX and Linux devices, and have found an issue when using autofs of NFS points for backup.

So this is the workflow:
1. Partition mount NFS points when backup is performed or when user access data on NFS, this is mounted on /nfs filesystem
2. Zabbix discovers new mount point and start monitoring it. Trigger is configured to for 15%(W), 10%(A) and 3%(H) and regular expression allows NFS shares, but since we have a lot of space on ZFS export, no trigger is lifted.
3. Now, the problem manifests itself when autofs unmounts this mount point. For some reason free disk percentage goes to 0% and trigger is lifted. I know that this mount point is now mounted on root filesystem, but what confuses me is that we have more than enough space left of root filesystem (80% free). This does not happen on our Linux machines.

I would be very grateful for any help and guidelines.

Update: zabbix-get and zabbix-agent -t return 0 as value on those filesystems:

root@host:/opt/zabbix/sbin# ./zabbix_agentd -t vfs.fs.size[/nfs,pfree]
vfs.fs.size[/nfs,pfree] [u|0]

[root@zabbix]# zabbix_get -I PUBLIC_IP -s host -k 'vfs.fs.size[/nfs,pfree]'
0.000000

So, we basically get wrong results back. We have tried to run zabbix agent as root, but the result is the same.

Best regards,
Dario

[tim.mooney]

I guess if I were in your situation, I would exclude /nfs from being monitored on the clients, since you're already monitoring the backend (NFS server exports) available free space.

[gopkris2000]

I guess if I were in your situation, I would exclude /nfs from being monitored on the clients, since you're already monitoring the backend (NFS server exports) available free space.

How did you excluded all the nfs mounts from the monitoring?

[tim.mooney]

Zabbix doesn't monitor anything that you don't tell it to monitor.

Are you using templates that auto-discovered the NFS mounts? If so, does your template have a method to skip (not include) certain mounts from the auto-discovery process?

[gopkris2000]

Zabbix doesn't monitor anything that you don't tell it to monitor.

Are you using templates that auto-discovered the NFS mounts? If so, does your template have a method to skip (not include) certain mounts from the auto-discovery process?

yes, I am using the Zabbix default "Linux OS template" which does not show any option to skip certain NFS mounts(the only option is needed to disable the trigger for each host), have tried ^nfs$ in regular expression but that is also not worked.

[tim.mooney]

Are you talking about this template:

https://git.zabbix.com/projects/ZBX/repos/zabbix/browse/templates/os/linux?at=refs%2Fheads%2Ffeature%2FZBXNEXT-6651-5.0

? If so, did you try setting the {$VFS.FS.FSTYPE.NOT_MATCHES} to the appropriate regex for NFS filesystem type before the discovery ?

You could try setting it now and then delete what was previously discovered for one of the clients, to see if it's re-discovered.