Constant disk activity

[krojew]

Hi,
Since last two updates, my FreeNAS system is having constant disk activity, as you can see in the attached image. The disk LED is constantly on. I don't really know if that's some kind of bug, or is it normal. Although I cannot remember such behavior in the past. Anyone has any ideas?

 

[joeschmuck]

Do you have any plugins/jails? Also please follow the forum rules and post the exact version of FreeNAS you are running.

I suspect thsi is normal, most of us have activity depending on what is running on our systems, also if the System Dataset is on the pool then you will have lots of activity as well.

 

[frankccc123]

Hi,
Since last two updates, my FreeNAS system is having constant disk activity, as you can see in the attached image. The disk LED is constantly on. I don't really know if that's some kind of bug, or is it normal. Although I cannot remember such behavior in the past. Anyone has any ideas?

same here, latest version, i had random lights stay on on the drive array, rebooted, different ones stay light

 

[krojew]

Do you have any plugins/jails? Also please follow the forum rules and post the exact version of FreeNAS you are running.

I suspect thsi is normal, most of us have activity depending on what is running on our systems, also if the System Dataset is on the pool then you will have lots of activity as well.

My version is FreeNAS-9.10.2-U2 (e1497f2). I have Plex and owncloud installed. When I disable them, the disk activity report shows no IO, but the activity LED is still on. That's quite strange. Also, I have system dataset on a usb drive.

EDIT: Sorry, my mistake - after disabling the plugins the activity is still on the report.

 

[joeschmuck]

My version is FreeNAS-9.10.2-U2 (e1497f2). I have Plex and owncloud installed. When I disable them, the disk activity report shows no IO, but the activity LED is still on. That's quite strange. Also, I have system dataset on a usb drive.

EDIT: Sorry, my mistake - after disabling the plugins the activity is still on the report.

I have my system dataset on my pool and Plex running and have the exact same small amount of activity and have had it for a very long time (looked back to Oct 2016). Are you sure your system dataset is on that USB Flash drive?

 

[krojew]

I have my system dataset on my pool and Plex running and have the exact same small amount of activity and have had it for a very long time (looked back to Oct 2016). Are you sure your system dataset is on that USB Flash drive?

Yes, I'm sure. Plex shouldn't be the issue since the activity persists without any plugins running. Could this indeed be a bug?

 

[styno]

To test this I installed an extra hdd (ada8) and moved the system dataset and reporting database to a pool that contains only this disk.
All other disks in the pool are now free of any iops other than the replication task that starts every hour. (no jails/no shares/...)
The other disks are now going into standby mode as well.

 

[krojew]

Since my system dataset is already not on any hdd, there's nothing to test in this regard for me. I'm wondering if there's some way to see what's actually causing this activity.

 

[styno]

Since my system dataset is already not on any hdd, there's nothing to test in this regard for me. I'm wondering if there's some way to see what's actually causing this activity.

The system dataset and reporting database are both moved? Theoretically you should be able to identify some processes with 'lsof' and 'fuser'.

 

[krojew]

Hmm I just checked to confirm and the system dataset was moved from my original usb stick for some reason. This is strange. But I guess this might cause the activity.

 

[SweetAndLow]

Constant disk activity is normal. It comes from jails, system dataset or rrd logs. There really isn't a way around it and it doesn't hurt anything. Letting your drives spin down will shorten their lifespan.

Sent from my Nexus 5X using Tapatalk

 

[styno]

Agreed, I mentioned the disks going to standby as additional prove that no iops were landing on them.

 

[FreeVel]

I am having the same issue however I have validated that it is not related to the system dataset. I can move it between datasets and I can explicitly trace it.

The issue only appears on only one of the three active jails. When I stop the suspicious jail, the constant disk I/O goes away. The jail runs a VPN server however even when the VPN server isn't running (incl. no other service) and as long as the jail is active the constant disk I/O of 100k carries on (~ 100k/sec, no spikes, just flat line of 100k/sec).

The jail is hosted on one of two raidz volumes with the second volume not affected by the constant disk I/O.

However, when I tried to aggregate my server's two network cards I noticed the disk I/O activity started affecting both Raidz volumes. When I remove the aggregate nic then the constant disk I/O activity remained in the jails' volume.

I cannot find what causes the VPN jail, even with no other services running, to cause constant disk I/O.

I suspect whether having an incorrect network setup, incl. the mixing of jails with and without VIMAGE enabled, could cause the issue.

Any ideas on troubleshooting are welcomed


UPDATE:
switching lz4 compression off on the volume seems to have cured the problem.