Anyone having problems recently with Areca 1224/1882 ?

[jimmyjoe]

I've had three disks go out on two different Proxmox servers in the past two days. Not sure what's up. Could be coincidence, maybe not. Just thought I'd check if anyone else is having problems due to kernel updates, patches, etc. Both boxes are fully patched running pve-manager/3.4-11/6502936f (running kernel: 2.6.32-43-pve)

On box #1 running Areca arc-1224:
I was restoring a VM from dump. About 50% through, Areca tossed one of the SSDs (Samsung EVO 1TB) from the RAID1 array. The box was completely locked up. Could not ssh in or login from console. Picture of the console messages is below. Weird that it failed that hard.... I rebooted (power button) and added a new hotspare.

The following day after the array was rebuilt, I tried the VM restore again. About 50% through and Areca tossed another disk from the same array. After adding *another* new disk, the array started rebuilding. I went home. We came in this morning and yet another disk was tossed from the array but the machine was responding. After the second disk went out yesterday, just for good measure, I updated the Areca firmware to latest 1.52 2014-12-26 and have rebooted.

Box #2 is running Areca 1882:
This morning it tossed a SATA disk from it's array. I've replaced that and things are quiet... for now.

This could all be wild coincidence, but generally speaking, these boxes have been running pretty much flawlessly for a couple years (with an occasional disk failure). Proxmox has been performing beyond expectation. Just wondering if anyone else is having problems too? I may boot to a previous kernel (and arcmsr driver) if it happens again.

 

[udo]

I've had three disks go out on two different Proxmox servers in the past two days. Not sure what's up. Could be coincidence, maybe not. Just thought I'd check if anyone else is having problems due to kernel updates, patches, etc. Both boxes are fully patched running pve-manager/3.4-11/6502936f (running kernel: 2.6.32-43-pve)

On box #1 running Areca arc-1224:
I was restoring a VM from dump. About 50% through, Areca tossed one of the SSDs (Samsung EVO 1TB) from the RAID1 array. The box was completely locked up. Could not ssh in or login from console. Picture of the console messages is below. Weird that it failed that hard.... I rebooted (power button) and added a new hotspare.

The following day after the array was rebuilt, I tried the VM restore again. About 50% through and Areca tossed another disk from the same array. After adding *another* new disk, the array started rebuilding. I went home. We came in this morning and yet another disk was tossed from the array but the machine was responding. After the second disk went out yesterday, just for good measure, I updated the Areca firmware to latest 1.52 2014-12-26 and have rebooted.

Box #2 is running Areca 1882:
This morning it tossed a SATA disk from it's array. I've replaced that and things are quiet... for now.

This could all be wild coincidence, but generally speaking, these boxes have been running pretty much flawlessly for a couple years (with an occasional disk failure). Proxmox has been performing beyond expectation. Just wondering if anyone else is having problems too? I may boot to a previous kernel (and arcmsr driver) if it happens again.


View attachment 3082

Hi,
I have no issues with my areca raids... but perhaps different kernels:

system 1+2: ARC-1222 2.6.32-41-pve
system 3  : ARC-1882 2.6.32-42-pve
system 4+5: ARC-1882 3.10.0-8-pve
system 6+7: ARC-1680 3.10.0-8-pve
system 8  : ARC-1222 2.6.32-37-pve
system 9  :  ARC-1882 3.10.0-10-pve

One SAS-drive is dying some days ago - but that's normal (some years old and heavy use - because backup spool disk).

But you should not use an Samsung EVO SSD in an raid... if you don't want trouble use an DC-grade SSD for this (e.g. Intel DC S3610).

Udo

 

[jimmyjoe]

Booted with 2.6.32-42-pve instead of -43 and all seems well now. Thanks for the heads-up on the EVO drives.

 

[jimmyjoe]

Booted with 2.6.32-42-pve instead of -43 and all seems well now. Thanks for the heads-up on the EVO drives.

Just an update - I don't think it had anything to do with switching kernels. I'm back to running -43 without any problems so far. I've swapped out the Samsung 850 drives for the older 840 model. We've run the 840 EVO and PRO on these controllers for years without problem. The Areca event log shows the 850s timing-out many times per day where the 840s only had two timeouts this year. As udo suggested though, we're upgrading to Intel S3510s. Nothing wrong here with Proxmox or the kernel.