Hi,

 

Good day! For months I did not have problem with my unraid setup till I decided to upgrade my 3x1TB cache to a single 500gb SSD and remove a drive (w/ errors) permanently from the array without replacing it.

 

After I replaced my cache with the SSD, I noticed everything became very slow. My VMs would freeze for 10-20seconds. I have reinstalled docker apps 3 times, while deleting the docker image 4 times and relocating the appdata once. Sometimes my webui would not be accessible and would require me to edit the config on my usb to not autostart the docker.

 

Also I removed a damaged 3tb HDD. I copy and pasted the files manually. Made sure that the drive was empty when I removed it and allowed the parity drive to rebuild.

 

On top of that, I have a 10gb port which I can transfer files (p2p 10gb) around 30+mbps (same speed on my 1gbps). Though the 10gb network starts at around 300-400 and stabilizes at 30+, while the 1gb network starts at 50-60 and stabilizes around 30+mbps.

 

How do I fix this?

sobnology-diagnostics-20171114-1757.zip

Constant syslog errors related to the SSD, possibly a bad cable:

 

Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: task abort: SUCCESS scmd(ffff88007d660600)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: attempting task abort! scmd(ffff88007d662400)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: [sdd] tag#14 CDB: opcode=0x28 28 00 06 b4 88 48 00 05 b8 00
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: handle(0x0009), sas_address(0x4433221100000000), phy(0)
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: enclosure_logical_id(0x500605b002c8b25c), slot(0)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: task abort: SUCCESS scmd(ffff88007d662400)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: attempting task abort! scmd(ffff88007d661b00)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: [sdd] tag#13 CDB: opcode=0x28 28 00 06 b4 88 00 00 00 38 00
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: handle(0x0009), sas_address(0x4433221100000000), phy(0)
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: enclosure_logical_id(0x500605b002c8b25c), slot(0)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: task abort: SUCCESS scmd(ffff88007d661b00)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: attempting task abort! scmd(ffff88007d662580)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: [sdd] tag#12 CDB: opcode=0x28 28 00 06 e1 0f 18 00 02 68 00
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: handle(0x0009), sas_address(0x4433221100000000), phy(0)
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: enclosure_logical_id(0x500605b002c8b25c), slot(0)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: task abort: SUCCESS scmd(ffff88007d662580)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: attempting task abort! scmd(ffff88007d661500)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: [sdd] tag#11 CDB: opcode=0x28 28 00 06 e1 0b 18 00 04 00 00
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: handle(0x0009), sas_address(0x4433221100000000), phy(0)
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: enclosure_logical_id(0x500605b002c8b25c), slot(0)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: task abort: SUCCESS scmd(ffff88007d661500)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: attempting task abort! scmd(ffff88007d661e00)
Nov 14 07:02:40 sobnology kernel: sd 17:0:0:0: [sdd] tag#10 CDB: opcode=0x28 28 00 06 e1 07 e0 00 03 38 00
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: handle(0x0009), sas_address(0x4433221100000000), phy(0)
Nov 14 07:02:40 sobnology kernel: scsi target17:0:0: enclosure_logical_id(0x500605b002c8b25c), slot(0)

 

 

P.S.: it would be best to move it to an onboard port since the LSI where is it connect won't be able to trim it.

Edited November 14, 20178 yr by johnnie.black

@johnnie.blackhey thanks for replying! The onboard ones are sata2, will that be fine? Should I move the parity drive there too?

I would prefer a trimmed SSD on SATA2 than an untrimmed one on SATA3, for parity it soen't matter which controller you use.

Also your board has SATA3 ports, just don't use the Marvell ports, the first four white ports.

@johnnie.black already switched them. Hope everythings gonna be ok now. How about the 10gb with 33mbps transfer speed?

1 minute ago, karlpox said:

How about the 10gb with 33mbps transfer speed?

 

Is that transferring to cache or the protected array?

@johnnie.blackcache (ssd) to ssd

Since the SSD wasn't being trimmed try again after running fstrim.

@johnnie.blackjust wondering if Im reading from the array (w/o cache) will the speed be just 30-40mbps? Im kinda having hard time streaming 4k content from the server to my nvidia shield.

16 minutes ago, karlpox said:

@johnnie.blackjust wondering if Im reading from the array (w/o cache) will the speed be just 30-40mbps? Im kinda having hard time streaming 4k content from the server to my nvidia shield.

 

No, it will be limited by the disk read speed where the data is, usually 100MB/s+ (I assume you mean 30-40MB/s, and not mbps).

30-40 MB/s (MegaByte/s) is more than enough for 4k video.

 

And that figure is when writing to the parity-protected drives, since the system then does a read/modify/write when updating the parity to avoid needing to read from all data disks to compute the new parity data.

 

Reads from the parity-protected drives are behaving as normal single-drive reads, so it's the read capacity of the drive and the network speed that will limit your transfer rates.

@johnnie.black yes sorry I meant 30-40MB/s. Im currently transferring files from unraid to my 2TB WD Black right now and im at 7-26MB/s. Also my parity check just completed

 

Last check completed on Wed 22 Nov 2017 02:13:44 AM SGT (today), finding 0 errors. 
Duration: 4 days, 9 hours, 19 minutes, 21 seconds. Average speed: 10.6 MB/sec

 

I used to Avg. 30+ MB/s here. Dont know what happened.

 

@pwm Sorry I was not clear. I am having 30-40MB/s on copy & paste with my P2P 10G connection. Im not sure what speed its running while streaming to my nvidia shield. Is there a way to find out?

 

 

Edited November 22, 20178 yr by karlpox

Post your current diagnostics, ideally showing the parity sync if you haven't rebooted since.

@johnnie.black

 

Using 10gbps I was able to transfer 300-400MB/s from server(SSD-cache) to my pc(SSD). Then I tried copying direct from the server(array) to pc(SSD) and the speed was around <26MB/s. Then I switched back to my 1gbps, the speed was reaching around 50MB/s sometimes but was more stable around 30+MB/s.

 

Lastly I tried streaming 4K content again with plex, and was able to watch a movie for an hour then the movie started freezing. I checked unraid and the disk usage was <15MB/s.

 

Is it possible to reach around 90-100MB/s transfer speed with the array?

sobnology-diagnostics-20171123-1424.zip

 

Im having 100MB/s write speed though.

Edited November 23, 20178 yr by karlpox

See if this helps with the read speed:

 

https://forums.lime-technology.com/topic/46802-faq-for-unraid-v6/?do=findComment&comment=526285