Hi everyone,

 

I'm experiencing some performance issues with my Unraid file server and could use some help diagnosing the problem.

 

I’m running an Intel i7-9700T CPU with 40GB of RAM and several hard drives, some as large as 18TB. Some of the drives are connected to the internal SATA ports, while others are attached to an LSI 9211-8i controller running in IT mode.

 

When resyncing the array, I get speeds of up to 200MB/s, which seems reasonable. However, when writing large files from one HDD to another within the system, the speeds drop to around 30-40MB/s, which feels unusually slow.

 

There also seems to be another issue that might be related to ZFS. Everything was working fine until I tried to stop the array. At that point, the system became almost unresponsive, with extremely high load and high I/O wait times. I have no idea what is causing this behavior.

 

I’ve attached diagnostics from the moment when I attempted to stop the array and everything went haywire. If anyone has insights or suggestions, I’d really appreciate your help.

 

Thanks in advance!

 

homeserver-diagnostics-20250320-2149.zip

Edited March 20, 20251 yr by TheDR

48 minutes ago, TheDR said:

However, when writing large files from one HDD to another within the system, the speeds drop to around 30-40MB/s, which feels unusually slow.

That's about normal for transfers between 2 array disks. Unlike writes from other sources the seeking needed to update parity is taxing.

Also ZFS in array is known to be slow.

Edited March 20, 20251 yr by Kilrah

Array to array writing is always going to be slow, but also be aware that's there's a known issue with zfs write performance for the array only:

 

https://forums.unraid.net/bug-reports/prereleases/zfs-write-speed-issue-array-disk-no-parity-on-6120-rc6-r2414/

 

5 hours ago, JorgeB said:

Array to array writing is always going to be slow, but also be aware that's there's a known issue with zfs write performance for the array only:

 

https://forums.unraid.net/bug-reports/prereleases/zfs-write-speed-issue-array-disk-no-parity-on-6120-rc6-r2414/

 

Ok, Is at the current state BTRFS better than ZFS? Because ZFS isn't useful for me, if it behaves this way while copying files from ZFS to Array. 

1 hour ago, TheDR said:

Is at the current state BTRFS better than ZFS?

For array usage, yes, since it doesn't have that write performance issue.

On 3/21/2025 at 4:45 PM, JorgeB said:

For array usage, yes, since it doesn't have that write performance issue.

Hmm... I think, now I really need some help.... Yes, the reading / writing from ZFS to the Array was slow, but now nothing works anymore.

 

The Pool was created, with Unraid itself. Then I copied all files from the Array to the Pool and replaced the Old drives. So I had to create a new Configuration. And I'm sure, that I configured the tool to just remove the Array Drives, also the ZFS Pool was gone in the configuration. But It's still there and was working. Now I can't access the Data on the ZFS Pool anymore. Even after the Upgrade to the 7.1 Beta with the new ZFS in the Kernel.

 

The Pool seems to be OK and imported:

root@HomeServer:~# zpool status -v
  pool: z1pool
 state: ONLINE
status: Some supported and requested features are not enabled on the pool.
        The pool can still be used, but some features are unavailable.
action: Enable all features using 'zpool upgrade'. Once this is done,
        the pool may no longer be accessible by software that does not support
        the features. See zpool-features(7) for details.
config:

        NAME        STATE     READ WRITE CKSUM
        z1pool      ONLINE       0     0     0
          raidz1-0  ONLINE       0     0     0
            sdg1    ONLINE       0     0     0
            sdk1    ONLINE       0     0     0
            sdh1    ONLINE       0     0     0
            sdj1    ONLINE       0     0     0

errors: No known data errors

 

The Filesystem seems also to be OK, but when I try to mount it, the whole System hangs:

 

root@HomeServer:~# zfs list
NAME     USED  AVAIL  REFER  MOUNTPOINT
z1pool  16.4T  12.1T  16.3T  /mnt/disks/z1pool
root@HomeServer:~# zfs mount z1pool

 

I get a load of about 10-15 with 50% IO Wait:

 

top - 17:20:30 up 21 min,  5 users,  load average: 13.32, 13.14, 8.79
Tasks: 409 total,   1 running, 408 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.0 us,  1.8 sy,  0.0 ni, 42.1 id, 56.0 wa,  0.0 hi,  0.0 si,  0.0 st

 

I also tried to scrub the drives. Got just a few Checksum Errors, that were corrected. But still no change. What's going on??

Do you mean you are importing the pool without mounting it? It should mount on import.

 

If that is the case, try importing read-only:

 

zpool import -o readonly=on z1pool