tomjrob Posted May 19, 2018 Share Posted May 19, 2018 I am upgrading my parity drive from 2TB to 4 TB. Problem is the throughput seems very low (35mb/sec to 44mb/sec). All drives in the array are 7200 rpm and a mix of sata 2 & sata 3. When doing preclear, the new 4TB drive was considerably faster (106mb/sec - 129mb/sec), so I am wondering if someone could shed some light on this. The drive is attached to same sata port for both the preclear and the parity rebuild (internal motherboard port on ASRock 970 Extreme 4) Diagnostics and preclear report are attached and I've uploaded hardware profile to Limetech as well. Thanks in advance for any insight. tower-diagnostics-20180519-1110.zip preclear_report_WOL240382382_2018.05.17_19.19.26.txt Quote Link to comment
tomjrob Posted May 19, 2018 Author Share Posted May 19, 2018 Additional Info: Screen shot of parity rebuild showing elapsed time Quote Link to comment
BobPhoenix Posted May 19, 2018 Share Posted May 19, 2018 (edited) It looks like you have a DVD/BluRay device installed. It also looks like you have it attached to your LSI card. I would hook it up via USB if possible or MB SATA port if not. It is possible that having it connected up to the LSI card is slowing down your parity check speeds. Log entry: May 18 09:47:03 Tower kernel: scsi 7:0:1:0: CD-ROM HL-DT-ST DVDRAM GH24NSB0 LN01 PQ: 0 ANSI: 0 May 18 09:47:03 Tower kernel: scsi 7:0:1:0: SATA: handle(0x000a), sas_addr(0x4433221106000000), phy(6), device_name(0x0000000000000000) May 18 09:47:03 Tower kernel: scsi 7:0:1:0: SATA: enclosure_logical_id(0x500605b002c8af85), slot(5) May 18 09:47:03 Tower kernel: scsi 7:0:1:0: atapi(y), ncq(n), asyn_notify(n), smart(n), fua(n), sw_preserve(n) May 18 09:47:03 Tower kernel: sr 7:0:1:0: [sr0] scsi-1 drive May 18 09:47:03 Tower kernel: cdrom: Uniform CD-ROM driver Revision: 3.20 May 18 09:47:03 Tower kernel: sr 7:0:1:0: Attached scsi CD-ROM sr0 May 18 09:47:03 Tower kernel: sr 7:0:1:0: Attached scsi generic sg7 type 5 I also see a port multiplier which would slow down a parity check - but the constant connects and disconnects below are the most likely (in my opinion) for your slow down: May 18 14:17:57 Tower kernel: ata6: exception Emask 0x10 SAct 0x0 SErr 0x40c0000 action 0xe frozen May 18 14:17:57 Tower kernel: ata6: irq_stat 0x00000040, connection status changed May 18 14:17:57 Tower kernel: ata6: SError: { CommWake 10B8B DevExch } May 18 14:17:57 Tower kernel: ata6: hard resetting link May 18 14:18:07 Tower kernel: ata6: softreset failed (device not ready) May 18 14:18:07 Tower kernel: ata6: hard resetting link May 18 14:18:15 Tower kernel: ata6: SATA link up 3.0 Gbps (SStatus 123 SControl 300) May 18 14:18:15 Tower kernel: ata6.15: Port Multiplier 1.2, 0x197b:0x0325 r0, 15 ports, feat 0x5/0xf May 18 14:18:15 Tower kernel: ata6.00: hard resetting link May 18 14:18:17 Tower kernel: ata6.00: failed to resume link (SControl 0) May 18 14:18:17 Tower kernel: ata6.00: SATA link down (SStatus 0 SControl 0) From the log ata6 has the following drives connected to the port multiplier - I would get another LSI card and stay way from the port multiplier: May 18 14:18:30 Tower kernel: ata6.01: ATA-8: Hitachi HUA723020ALA640, MK0271YGJ7RVHA, MK7OA800, max UDMA/133 May 18 14:18:30 Tower kernel: ata6.01: 3907029168 sectors, multi 0: LBA48 NCQ (depth 31/32), AA May 18 14:18:30 Tower kernel: ata6.01: configured for UDMA/133 May 18 14:18:30 Tower kernel: ata6.02: ATA-8: ST2000DM001-1CH164, W1E3Y8ZF, CC26, max UDMA/133 May 18 14:18:30 Tower kernel: ata6.02: 3907029168 sectors, multi 0: LBA48 NCQ (depth 31/32), AA May 18 14:18:30 Tower kernel: ata6.02: configured for UDMA/133 Maybe swap the DVD and the Hitachi above. Then find another port on the LSI (or another LSI card) for the Seagate. Then the port multiplier would only be servicing your DVD drive where it hopefully won't hurt anything. Edited May 19, 2018 by BobPhoenix Quote Link to comment
tomjrob Posted May 19, 2018 Author Share Posted May 19, 2018 Bob, Thanks for the quick reply. I do have a dvd drive attached to the lsi card. I believe the "port multiplier" you are referring to is an external drive bay (Mobilestor unit) which is attached via esata to the unraid server. I only use it for "unassigned devices" which are not part of the parity protection devices, so didn't think that would cause issues. However, if it is causing resets, I will remove it. So the plan is to disconnect the DVD drive, and turn the Mobilestor unit off and run a parity check and compare performance. Does that make sense? Quote Link to comment
BobPhoenix Posted May 19, 2018 Share Posted May 19, 2018 Yes. Hopefully that will take care of you problem. 35-44 isn't bad when writing to your array but I think you should be getting more like 50 maybe even 60. Another option is to turn on TurboWrite mode (I believe that is "reconstruct write") in settings. That should improve your write speeds at the expense of having to spin up ALL drives when doing a write to the array. Quote Link to comment
BobPhoenix Posted May 19, 2018 Share Posted May 19, 2018 (edited) 3 hours ago, tomjrob said: Bob, Thanks for the quick reply. I do have a dvd drive attached to the lsi card. I believe the "port multiplier" you are referring to is an external drive bay (Mobilestor unit) which is attached via esata to the unraid server. I only use it for "unassigned devices" which are not part of the parity protection devices, so didn't think that would cause issues. However, if it is causing resets, I will remove it. So the plan is to disconnect the DVD drive, and turn the Mobilestor unit off and run a parity check and compare performance. Does that make sense? You are correct in that Unassigned devices should NOT cause problems when writing to the array. The disconnects probably are not causing you any real problems then but it won't hurt to test with it disconnected to make sure. So if it isn't the DVD drive on the LSI or the disconnects then I would turn on the Turbo Writes and see if that gets you a significant increase in performance. I had some drives that were just SLOW. They had NO smart problems but when I took them out of my array the speed went up dramatically. So you may have a drive that is developing slow sectors but are not producing smart errors yet. You might want to test your drives for that and see if you can find any slow spots. Then replace any disks that are slow either because they are older (newer drives are usually faster than older drives) or they are developing problems that smart hasn't reported yet. That is all I've got for you at this point maybe someone else has additional/better ideas. Edited May 19, 2018 by BobPhoenix Quote Link to comment
tomjrob Posted May 20, 2018 Author Share Posted May 20, 2018 Parity check speed was 107.9MB/sec after disconnecting DVD drive. The Mobilstor external esata device did not seem to affect the speed, as I turned it off and on during the check and speed did not change significantly. I will isolate or remove the DVD drive going forward. Much thanks to BobPhoenix for the assistance. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.