3 Drives Showing High Read Errors

[jehud]

I woke up today and my server was unresponsive and unreachable on the network. I had to hard reset it, and when it came back up three of the disks have a ton of read errors. Monday is parity running day which usually takes a day or two and is running now which is just increasing the read errors on those drives. I have attached the diagnostics dump from my server. What process should I follow, I'm guessing the disks may actually be fine and I just need to spin them down and then up or rebuild them. What would be the guidance here?

diagnostics-20200615-1110.zip

[trurl]

6 minutes ago, jehud said:

Monday is parity running day which usually takes a day or two

Are you saying you do weekly parity checks? Why? Parity is maintained realtime, every time an array disk is written, parity is updated. Parity check is just for confirming that parity is indeed valid, which it should be unless something is wrong. Parity checks are not required to maintain parity. Spending 1 or 2 out of 7 days checking parity seems excessive. Most only do monthly parity checks.

 

Disks 4,5,6 have disconnected. Shutdown, check all connections, power and SATA, both ends, including power splitters. Then post new diagnostics.

 

Are these disks on the Marvell ports?

[jehud]

There is a good chance they are on Marvell ports yes (l have an expansion card in there), I had to do some hack like work around a while back after one of the OS updates to get the drives showing up again which got broken on a release. Everything has been pretty much fine since then though.

 

Good tip on the parity it did seem to be excessive in how I was running it I probably should have checked the docs out.

 

Let me try to do those checks shortly and get another set of data up.

[jehud]

I rebooted again and everything is fine now, do you want me to include the diagnostics anyhow?

[JorgeB]

4 hours ago, trurl said:

Are these disks on the Marvell ports?

They are, and the most likely problem.