Resource icon

Building, Burn-In, and Testing your FreeNAS system

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
Any chance you might add support for nvme disks?
I only see my DOMs.

It's a tool that was written back when PCI (not PCIe, but *PCI*) was relatively new. Give it a break. ;-)

That said, it should still be able to test your devices if you give the device names (see option 3). I know FreeBSD was making some progress towards some sort of better subsystem for NVMe, similiar to CAM for SCSI/SATA, but I don't know what the state of that is. Do your devices show up as... what? Is it nvdX? I don't remember offhand but give option 3) a try with the device names of all your NVMe disks and see if that works, that'll give some useful information on how easy and useful this would be.
 

ehsab

Dabbler
Joined
Aug 2, 2020
Messages
45
It's a tool that was written back when PCI (not PCIe, but *PCI*) was relatively new. Give it a break. ;-)

That said, it should still be able to test your devices if you give the device names (see option 3). I know FreeBSD was making some progress towards some sort of better subsystem for NVMe, similiar to CAM for SCSI/SATA, but I don't know what the state of that is. Do your devices show up as... what? Is it nvdX? I don't remember offhand but give option 3) a try with the device names of all your NVMe disks and see if that works, that'll give some useful information on how easy and useful this would be.
It sure does work by specify the device names. So the test is running atm :)
Thanks for taking the time to reply.
 

ehsab

Dabbler
Joined
Aug 2, 2020
Messages
45
@jgreco the disks i tested above was Samsungs PM983 which is a PCIe gen3 x4 NVMe disk.
I now have a PM1733 disk which is a PCIe gen4 x4 with theoretical two times the performance of the PM983.
But for some reason the same test just gives me ~1.4GB/s, any ideas why this could be?
 

sindreruud

Dabbler
Joined
Sep 21, 2017
Messages
24
Looks like a really useful script, thank you!

I have a question regarding splitting up the test. I am receiving two 18tb drives in a couple of days, and plan on using this script. However, my apartment is tiny and the mini-server is in my living room. I am assuming this script will make the harddrives create quite a lot of noise non-stop for however many days it take?

So in an ideal world I would start this test before going to bed and then "pause" it when i wake up, or when i get back from the office. Is this at all possible or necessary?
What about just stopping the test and then restarting it a couple of hours later, will that be as effective in testing the drives? I'm afraid stopping the test and then starting a new one will make me miss out whatever it tests towards the end.

Edit: Sorry, I've seem to have posted in the wrong thread. Should probably have put this in the solnet discussion thread: https://www.truenas.com/community/threads/solnet-array-test-discussion-thread.57317/page-3

My bad.
 
Last edited:

nutrient2314

Cadet
Joined
Jan 26, 2024
Messages
6
I can understand that burn-in is very important to rule out drive failures in future, however, I wonder, what possible amount of burn-in actually makes the drive less resilience in production? In other words, do you think burn-in will actually reduce the life of the pool? I know SSD will report a significant drop in TBW, what about rust HDD?
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
SSDs are a tricky subject, but for HDDs the answer is straightforward - there should be no meaningful degradation. Yes, the mechanical bits will wear out sooner or later, but it's not remotely comparable.
 
Top