Scale and core keep my disks busy.

Rexodus

Cadet
Joined
Jan 7, 2023
Messages
5
Hi there!

Running up to date versions of Scale and Core in an up to date and flawlessly working Proxmox, I hear both Core and Scale do the same. They keep my RAID5 ZFS array busy and I like the 4 data-disks to go to sleep if not accessed for days. Because them both show this behavior, it seems like a feature but I'm not sure. If it is a feature, I like to switch it off. But I have no clue what is keeping them busy.

For completeness I've created a sound-file. So you can hear what I hear. The first 20 seconds, you can hear only the fan. Truenas Core is booting. After about 20 seconds, it finds the array. You can hear lots of noises, like expected. But after the boot is complete, it keeps making noises. It never stops. Nothing or nobody is logged in. Nothing should access the disks.

In the space of about a year, I had several re-installs. But this is persistent.

I googled over the time. But I can't find a similar problem. Something like diskmon and filemon in DOS back in the day would be nice. But I have no clue what tool does things like that in BSD/Linux.

Anyway, any clue would be appreciated.
 

Davvo

MVP
Joined
Jul 12, 2022
Messages
3,222
my RAID5 ZFS array

I don't know a thing about virtualization but I believe you want to search for hdd spindown. There is a script in the resource section.
 
Last edited:

Rexodus

Cadet
Joined
Jan 7, 2023
Messages
5

I don't know a thing about virtualization but I believe you want to search for hdd spindown. There is a script in the resource section.

Why would a spindown script work if something keeps accessing the 'not relevant what file-system it contains' disks?
 

Davvo

MVP
Joined
Jul 12, 2022
Messages
3,222
Why would a spindown script work if something keeps accessing the 'not relevant what file-system it contains' disks?
Because from the message it wasn't clear if you were aware of the deafult behaviour of HDDs and it looked like you were expecting they would automatically spin down.
Did you move the pool's dataset out of the HDDs?
 

Rexodus

Cadet
Joined
Jan 7, 2023
Messages
5
Did you move the pool's dataset out of the HDDs?

I can delete them. Which is an option I don't look forward to. At least, if I understand what you try to say. A restore will cost me days. Moving is no option either. I simply don't have the storage.

I need a tool like WireShark for HDD's. That would help a lot. But I can't find anything. It doesn't exist or I can't search good enough...
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,680
need a tool like WireShark for HDD's.

Running tcpdump to observe the network activity isn't sufficient?

zpool iostat should help provide some guidance as to whether it is read or write activity.

fstat can show you what processes have what file resources open (you might need to read the man page to identify the best set of flags to use).

If you're really desperate, the audit subsystem (audit, auditd, etc) is capable of giving you super-fine-grain inspection capabilities as to what I/O is going on. TrueNAS technically doesn't support this but last time I checked some years ago, all the bits seemed to be there. That really would be close to a "WireShark for HDD's".
 

Davvo

MVP
Joined
Jul 12, 2022
Messages
3,222
I can delete them. Which is an option I don't look forward to. At least, if I understand what you try to say. A restore will cost me days. Moving is no option either. I simply don't have the storage.
I meant this.
 

Rexodus

Cadet
Joined
Jan 7, 2023
Messages
5
I meant this.
It's silent now \o/.

Thank you very much! I looked at it a while ago but I didn't have the balls to do that. Expecting the whole thing to go up in smoke.

Anyway, now my 11 years old disks can go to sleep. They will be tired! The absence of any sound means they thank you too!

One large statue for you, Davvo!

Greetings, Rex
 
Top