Reads every 5 minutes, but not on core

FlangeMonkey

Contributor
Joined
Dec 6, 2012
Messages
111
Hi Guys,

Usually the case, when you start looking at stuff in detail, you find shi-uff that might be normal.

I've moved the system datasets to my boot pool and no ix-application datasets so the system should be idle, however, I see small reads every 5 minutes:

1614265529245.png


For testing's sack as this is a new pool and new disks, I imported it into core and core doesn't do this:
1614272396847.png


Does anyone know what this could be? I don't see anything in JIRA or other reports.

Thanks,
 
Joined
Feb 27, 2021
Messages
5
Just wanted to add that I'm having the same issue. I'm getting reads every 5 minutes on my HDD, even after my ix-application datasets and bootpool and system pool have been moved to a separate SSD. I've been trying to track down the process doing it with IOTOP but I'm struggling to pinpoint it.

1614447631314.png
 

shadofall

Contributor
Joined
Jun 2, 2020
Messages
100
USB or Sata or other?
SSD or HDD
and are the disk assigned to any pool? i had one disk doing that a SATA M.2 in a USB Chassis that i had plugged in doing nothing. not assigned to anything.
 
Joined
Feb 27, 2021
Messages
5
USB or Sata or other?
SSD or HDD
and are the disk assigned to any pool? i had one disk doing that a SATA M.2 in a USB Chassis that i had plugged in doing nothing. not assigned to anything.

In my case, it's a SATA HDD that is assigned to a single pool. I don't have any data stored on the pool yet. I am still setting things up.

I tried switching off S.M.A.R.T. under the disk options and it stops the every 5 minutes reads, but it checks it every 30 minutes (wakes the drive up and I see reads getting reported) and I get a notification that S.M.A.R.T isn't functioning on the drive. (Makes sense because I shut it off to test.)

I am trying to figure out how to not have SMARTCTL wake up the drive every 5 minutes. At least, I'm assuming that it's SMARTCTL that is causing it here.
 

FlangeMonkey

Contributor
Joined
Dec 6, 2012
Messages
111
My analysis also points to smart, as I get the same results if I execute a "smartctl --all". However, it does this every 5 minutes with the smart service running or not.

Edit: all disks are SATA HDD, single RAIDZ2 vdev and pool. However different results between Core and Scale.
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
In my case, it's a SATA HDD that is assigned to a single pool. I don't have any data stored on the pool yet. I am still setting things up.

I tried switching off S.M.A.R.T. under the disk options and it stops the every 5 minutes reads, but it checks it every 30 minutes (wakes the drive up and I see reads getting reported) and I get a notification that S.M.A.R.T isn't functioning on the drive. (Makes sense because I shut it off to test.)

I am trying to figure out how to not have SMARTCTL wake up the drive every 5 minutes. At least, I'm assuming that it's SMARTCTL that is causing it here.
Okey you should file a JIRA bug report about this. Because SMARTCTL shouldn't wake drives every x minutes regardless of status.
 

Hisenmisen

Cadet
Joined
Oct 2, 2022
Messages
1
I do not know anything about this, but couldn't this be the ZFS RAM cache being updated? As said I dont know much about these sorts of things so don't trust what I'm saying. Could someone who is more knownledgeable than me confirm or deny my theory? I would very much want to know what these spikes are as I also have them.
 

Jimi_M

Cadet
Joined
Nov 22, 2023
Messages
1
I need to elaborate, I have the same issue: every 5 minutes, there's disk activity preventing it from spin down. The dataset is on the boot pool, no plugins are installed, all services are off. This is a home server used once every few days, so it'd be best if the disks could stay in standby mode or shut down when not in use, to extend their lifespan. Is it possible that such a simple thing is causing so many problems?"
 
Top