Hello,

 

I'm having an odd issue that appears to only be happening with one disk. When I copy a file from my PC over LAN the files copy extremely slow, they start out at a decent speed and then drop down to KB speeds. When playing high bitrate videos (30+ Mb/s) I'm getting constant buffering every few minutes. The disk is a Deskstar 4TB drive and I have another disk that is the same and does not have this problem. I have 10 disks in the array including parity. I have been attempting to copy some files from disk5 to disk8 (problematic disk) and Windows Explorer completely locks up and it can't even copy. I have no issues with any other disk in terms of copying/playing files so I thought maybe it was a disk problem or cable issue. I swapped out the SATA cable and even switched the port that it was plugged into. I ran short/full SMART tests on the disk and received no errors. I ran the command to test read/write speeds on all disks and every disk, including the problematic disk had normal read/write speeds (oddly enough the problematic disk had one of the highest read/write). I'm on 5.0.5 right now if that matters. I thought maybe there was an issue with not enough power since I recently installed two new disks so I spun down all the disk and spun up disk8 and still saw the same poor performance. If I'm playing a file from disk8 at high bitrate and I start to copy a file to disk8 I get buffering almost immediately. This does not happen with any other disk, and I don't think it should - writing and reading should not effect one another. I think the disk should be able to write at <40Mb/s and read at the same time without interrupt.

 

My log file doesn't show any errors but I did notice an odd spin down reporting where the same disks are being called to spin down just a few moments after it has already been called to spin down. I turned spin down off for disk8 because I thought maybe this was the problem, still have the same issues.

 

Dec 12 09:04:38 TAYLOR-NAS kernel: mdcmd (111): spindown 3

Dec 12 09:04:40 TAYLOR-NAS kernel: mdcmd (112): spindown 4

Dec 12 09:05:01 TAYLOR-NAS kernel: mdcmd (113): spindown 1

Dec 12 09:05:01 TAYLOR-NAS kernel: mdcmd (114): spindown 2

Dec 12 09:05:03 TAYLOR-NAS kernel: mdcmd (115): spindown 5

Dec 12 09:05:04 TAYLOR-NAS kernel: mdcmd (116): spindown 6

Dec 12 09:05:05 TAYLOR-NAS kernel: mdcmd (117): spindown 9

Dec 12 13:04:29 TAYLOR-NAS kernel: mdcmd (118): spindown 1

Dec 12 13:04:30 TAYLOR-NAS kernel: mdcmd (119): spindown 3

Dec 12 13:04:32 TAYLOR-NAS kernel: mdcmd (120): spindown 9

Dec 12 13:05:02 TAYLOR-NAS kernel: mdcmd (121): spindown 2

Dec 12 13:05:04 TAYLOR-NAS kernel: mdcmd (122): spindown 4

Dec 12 13:05:05 TAYLOR-NAS kernel: mdcmd (123): spindown 5

Dec 12 13:05:05 TAYLOR-NAS kernel: mdcmd (124): spindown 6

Dec 12 13:05:06 TAYLOR-NAS kernel: mdcmd (125): spindown 7

Dec 12 17:22:31 TAYLOR-NAS kernel: mdcmd (126): spindown 4

Dec 12 17:22:32 TAYLOR-NAS kernel: mdcmd (127): spindown 5

Dec 12 17:27:22 TAYLOR-NAS kernel: mdcmd (128): spindown 3

Dec 12 17:28:24 TAYLOR-NAS kernel: mdcmd (129): spindown 1

Dec 12 17:28:35 TAYLOR-NAS kernel: mdcmd (130): spindown 2

Dec 12 17:28:36 TAYLOR-NAS kernel: mdcmd (131): spindown 6

 

Any help is appreciated, I can't stand having this issue considering I have 3.5TB of video files on this disk and they are almost all high bitrate making them basically unwatchable.

 

Edit: The disk spin down was interpreted wrong on my part, I wasn't reading the timing correctly. I run a scheduled job that moves local files onto the server every few hours so that would explain the spin downs occurring frequently.

So I decided to pull the drive and put it back in to rebuild the data, still have the same problem. Any help is really appreciated, I have tried everything I can think of so far.

This is often the precursor to failure, I would do some SMART checks on the drive

Are there other SMART tests to run? I already ran a long SMART test and received no errors. I do think it is probably a disk issue, but without any errors I'm not sure how else to tell.

Please provide a copy of the SMART report, and attach a current syslog, preferably zipped, that covers a period of slow performance of the drive, such as when it appears to be buffering.  Please see the first 3 posts of this thread.

SMART Test and System Log attached, I have been seeing the same issue this evening and previous few days so it occurred during the time period where the system log would pick it up.

 

unRaid OS 5.0.5

 

Server Specs:

Cooler Master CM Storm Series Trooper Case

Cooler Master V700 700W 12V PSU

3 x Icy Dock MB154SP-B 4x3 SATA 6gbps Swap Cage

MSI 970A-G46 AM3+ AMD 970

AMD Athlon II X2 270 CPU

G. Skill 8GB DDR3 SDRAM

 

Thanks! Sorry for not posting this info earlier.

smart01.txt

syslog.zip

I'm sorry I couldn't help, there are no issues at all with Disk 8.  There is one brief flurry of interface issues to the Parity drive, on Dec 10 between 12:33pm and 12:39pm, but that is the only drive issue visible.  So technically there are 3 periods when performance would have been affected.

 

1. During the interface issues (unknown cause, power?) period of almost 6 minutes on Dec 10 12:33pm - writes would have been slower and streaming might have stuttered.

2. During the 5 minutes the array was up with Disk 8 unassigned - performance to the simulated Disk 8 would have obviously been very poor.

3. During the almost 13 hours that Disk 8 was being rebuilt - performance to the simulated Disk 8 would have obviously been very poor.

 

Except for possibly during the first, I'm sure you weren't testing during the others, so I'm sorry I'm out of ideas.  Disk 8 looks as good as the rest.  Only the parity drive had a problem, but only during a 6 minute window, about 10 minutes after booting.  After it dropped the SATA link speed from 6gbps to 3gbps, then no more issues, and I don't think that would have caused much of an issue, 3gbps is fast enough and not a bottleneck.  I'm not sure what to conclude about that.  You might switch the parity drive to another port, and switch the cable, but there's no direct evidence of a bad cable or bad port.

Thanks for the update, I think what I'm going to do is pull the drive,format it and re-insert to rebuild from a blank state.