Slow Parity Rebuild

[JohnnyT]

I had a drive go bad "disk 5" and swapped it with a new one. The drive I swapped was working fine in a windows box before I put it in and all the drives show passed for smart test. The issue is that when I do a parity rebuild it is very very slow 162.8 KB/sec. I unplugged all the cables and plugged them back in. I also did a reboot. I am not sure what the issue is. Any help would be awesome.

diagnostics-20190207-2223.zip

[JorgeB]

Try replacing the SATA cable on the parity disk, but it's connected to a controller with a port multiplier, and those are a known issue.

[JohnnyT]

I will try that when I get home. I had tried moving the new drive to the board itself instead of a card with a new sata cable and that did not change anything. Thanks

[JohnnyT]

I was looking to get this "LSI 9201-16e External SAS HBA" and put most of my drives on that. Would that be a lot better than what I have now? I know there is a firmware update to do what I need.

[JorgeB]

6 minutes ago, JohnnyT said:

Would that be a lot better than what I have now?

Much better, LSI are currently the most recommended HBAs

[JohnnyT]

I will order it and then go from there. Thanks

Edited February 8, 2019 by JohnnyT

[JohnnyT]

I got the cables swapped and it is going much faster now 95 MB/sec but I am having another issue. My docker service won't start "Docker Service failed to start." This is what I am seeing in the log "root: truncate: cannot open '/mnt/cache/system/docker/docker.img' for writing: Read-only file system" Any ideas?

[JohnnyT]

After some more digging it looks like the cache pool is set to read only. I am looking how to fix that. It is a raid 10 setup.

[ezhik]

You also want to verify whether or not your write cache is enabled on your drives:

--

# hdparm -W /dev/sd[a-z]

 

Sample output:

 

/dev/sdw:
 write-caching =  1 (on)
 

--

# hdparm -I /dev/sd[a-z] |egrep -i 'serial number|model|write cache|/dev/'

 

Sample output:

 

/dev/sdw:
    Model Number:       HGST HUS724040ALA640                    
    Serial Number:      PN2356P6A9FMCC
       *    Write cache

 

NOTE: The star next to 'Write cache' indicates it is enabled. 

Edited February 9, 2019 by ezhik

[JorgeB]

3 hours ago, JohnnyT said:

Any ideas?

 

Please post current diagnostics.

[JohnnyT]

Here are the current diagnostics.

diagnostics-20190209-0925.zip

[JorgeB]

Cache pool is kind of a mess, of have multiple devices with errors:

 

Feb  8 17:43:54 davidFlix kernel: BTRFS info (device sdp1): bdev (null) errs: wr 5498, rd 974, flush 427, corrupt 0, gen 0
Feb  8 17:43:54 davidFlix kernel: BTRFS info (device sdp1): bdev /dev/sdo1 errs: wr 576481, rd 503357, flush 5101, corrupt 0, gen 0
Feb  8 17:43:54 davidFlix kernel: BTRFS info (device sdp1): bdev /dev/sdj1 errs: wr 0, rd 347, flush 0, corrupt 0, gen 0

These are hardware errors, these could be cable related, especially cache3, since SMART looks fine, cache1 shows some issues, but start by replacing cables also.

 

Then I would recommend you backup any important data on cache to the array and reformat the pool, after keep an eye on the btrfs device stats, see here for more info:

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=700582

 

[JohnnyT]

Thanks, I have been planning on redoing it anyways so now it's likely a good time.

[JohnnyT]

I ended up just blowing away the cache and starting over. My array seems to be all good, Thanks for the help