During the last few months, parity check has be leading to my server completely freezing (not ever reachable by ssh). The only way to regain access is a hard reset.

Some details:

• The server will rub for days or weeks and only hang during a parity check. Transferring lots of big files is not a problem.

• I was running unraid 6.12.15 but updated to 7.2.3. That did not fix the problem.

• The parity check will run for at least the first 8 TBs and but freeze some time after that (my server is 14TBs, a parity drive is 18TBs - although I do not think this could be relevant)

• I have not changed any hardware

• VMs and docker are disabled.

• I had enabled Mirror syslog to flash so I am attaching it but I cannot see anything that would cause the problem

• When I did a parity swap some months ago, I had to try multiple times (maybe 5 or 6) unit I succeeded (this was quiet anxiety inducing)
  

Any help would be welcome. I am expecting a new HDD any day now and I cannot replace my parity while this problem is going on.

syslog-previous-2026.02.24

Attach Diagnostics ZIP to your NEXT post in this thread.

Thanks (and I am attaching them here).

towers-diagnostics-20260225-0027.zip

Edited February 24Feb 24 by papnikol

Also

Feb 20 13:18:24 towerS root: Fix Common Problems: Warning: Marvel Hard Drive Controller Installed

Thank you both for taking the time.

8 hours ago, trurl said:

But my CPU is not Ryzen. It is indeed AMD but pre-Ryzen - Athlon X4 840, released in the ancient times of 2014
Also, would it start being problematic after so many years (regarding this aspect)?

8 hours ago, trurl said:

Also

Feb 20 13:18:24 towerS root: Fix Common Problems: Warning: Marvel Hard Drive Controller Installed

I am aware of that warning but it was not a problem for me for many years (apart from a few parity errors). As I said before, Why would it suddenly start being problematic now?

If the server is hanging during high load, it may have devolped a hardware issue, worth checking the CPU temps, if you have multiple RAM sticks, also worth trying with one at a time.

The interesting (and perplexing for me) part is that parity check works ok while 2 parity drives and 11 data drives are being read.
It usually freezes after passing the 8TB mark, meaning that 7 of the data drives do not have to be read anymore and spin down.

For load that won't matter, and it may be related more to the time it's been running.

14 minutes ago, JorgeB said:

For load that won't matter, and it may be related more to the time it's been running.

Well, it will mean less bandwidth and heat. I also tried pausing parity check every few hours (just a hail mary) and it did not help.

One thing I wanna try is rerun memtest (I ran it less than a year ago but you never know) but I have no monitor currently available and i have not found a way to run it headless.

There is a plugin for testing memory on a live system. It can only test unused memory but it can still be worth trying as if it produces any failures you know you have a RAM issue.

53 minutes ago, papnikol said:

Well, it will mean less bandwidth and heat.

Not for the CPU and RAM, with dual parity, the calculations still need to account for the disks that passed their capacity.

5 hours ago, papnikol said:

I am aware of that warning but it was not a problem for me for many years (apart from a few parity errors). As I said before, Why would it suddenly start being problematic now?

Marvell is notorious for not keeping their Linux drivers updated. Maybe it worked OK on an older version of Unraid but doesn't work so well on newer versions.

Your sig says you are using Unraid 5.0.5 by the way, which is obviously not true. Click on your username/icon at the top of the forum and select Account Settings. There you can enable sig viewing and edit your sig.