FreeNAS reboots by itself upon single disk failure

[william18]

Hi, my current FreeNAS build is FreeNAS-9.10.1-U4 (ec9a7d3). It is a Lenovo X3650 server with (11 x 4TB SATA hard disk) and (1 x 960GB SSD).

The single SSD drive is used as ZIL. The boot disk is using one of the 11 hard disk. The remaining 10 hard disks is configured as a single VDEV (Raid-Z3).

I am using it primarily as a datastore for VMWare.

The problem that i am facing is that on occasions when a single hard disk of the VDEV fails, FreeNAS will reboot by itself.

This should not be the case as FreeNAS should be able to handle a single hard disk failure gracefully?

Have anyone encountered such issues before? The output of my storage adapter is below. Any advice will be helpful to me. Thanks in advance!

Code:

# mfiutil show adapter
mfi0 Adapter:
	Product Name: ServeRAID M5210
		Firmware: 24.12.0-0024
	 RAID Levels: JBOD, RAID0, RAID1, RAID10
  Battery Backup: not present
		   NVRAM: 32K
  Onboard Memory: 0M
  Minimum Stripe: 64K
  Maximum Stripe: 64K

# mfiutil show drives
mfi0 Physical Drives:
 1 (  894G) JBOD <MZ7LM960HCHP-000 CB30 > SATA E1:S0
 2 ( 3726G) JBOD <ST4000NM0033 LB5F > SATA E1:S2
 3 ( 3726G) JBOD <ST4000NM0033 LB5E > SATA E1:S11
 4 ( 3726G) JBOD <MG04ACA400N TK42 > SATA E1:S4
 5 ( 3726G) JBOD <ST4000NM0033 LB5F > SATA E1:S8
 6 ( 3726G) JBOD <ST4000NM0035 LJ84 > SATA E1:S3
 7 ( 3726G) JBOD <ST4000NM0033 LB5F > SATA E1:S6
 8 ( 3726G) JBOD <ST4000NM0033 LB5F > SATA E1:S9
 9 ( 3726G) JBOD <ST4000NM0033 LB5F > SATA E1:S10
10 ( 3726G) JBOD <ST4000NM0035 LJ84 > SATA E1:S5
11 ( 3726G) JBOD <ST4000NM0033 LB5F > SATA E1:S7
12 ( 3726G) JBOD <ST4000NM0033 LB5E > SATA E1:S1

 

[Chris Moore]

This is a known issue it is because of swap space use on the drive. The problem is fixed in version 11.1 and all you need to do is upgrade.

 

[danb35]

The boot disk is using one of the 11 hard disk.

Why on earth would you waste a 4 TB disk as a boot device?

 

[william18]

This is a known issue it is because of swap space use on the drive. The problem is fixed in version 11.1 and all you need to do is upgrade.

Thanks a lot!

After reading your advice, I did some search and managed to find this post which talks about how to relocate the swap to another device/partition: https://forums.freenas.org/index.php?resources/how-to-relocate-swap-to-an-ssd-or-other-partition.68/

I am thinking to give this a try first.

 

[Chris Moore]

Thanks a lot!

After reading your advice, I did some search and managed to find this post which talks about how to relocate the swap to another device/partition: https://forums.freenas.org/index.php?resources/how-to-relocate-swap-to-an-ssd-or-other-partition.68/

I am thinking to give this a try first.

That's a workaround that was developed when there was no fix. The fix is in now, why not upgrade to the fixed version instead?

Sent from my SAMSUNG-SGH-I537 using Tapatalk

 

[danb35]

The fix is in now, why not upgrade to the fixed version instead?

Because the fixed version breaks other things for many people. Though I guess it's worth a try; reverting to the previous boot environment is straightforward enough.

 

[wblock]

A 960G SSD used for SLOG is overprovisioned by about 600%. ZFS only uses 16GB of SLOG space. Do not partition it and use it for several SLOGs, as that defeats the purpose. If you really need SLOG (many do not), one or more smaller, low-latency SSDs are faster.

 

[Chris Moore]

Because the fixed version breaks other things for many people.

My installation isn't very sophisticated so it hasn't caused me any trouble. Other than just one unauthorized reboot.

 

[Chris Moore]

The problem that i am facing is that on occasions when a single hard disk of the VDEV fails, FreeNAS will reboot by itself.

This should not be the case as FreeNAS should be able to handle a single hard disk failure gracefully?

Have anyone encountered such issues before?

One of the reasons you are having the unexpected / catastrophic hard drive failures is that hardware RAID controller you are using. That is not the right hardware for the job.