methanoid Posted January 31, 2016 Share Posted January 31, 2016 Been running for under 2 weeks but noticed I had some disk reallocations on a disk. So started to move data off it as I had space. Then unRAID decided to tell me it was over threshold after I think I got all the data off. Then I got more reallocations on further disks without capacity to move data off. These just seemed to be growing! So, ordering new disks (these are 6 weeks old only!) and wanted to look at the first fault one. Then 4 dropped out!! Opened PC, disconnect/reconnect cables and they reappear but Disk 1 - 160 reallocated Disk 2 - 176 Disk 3 - 136 (and red crossed out of action) Disk 4 - 96 All are on a molex to 4 SATA adapter - the other disks 5,6,7,Parity all zero errors on a different adapter. So, do I do a XFS_repair on Disk 3 now or wait until I have 4 replacement drives and swap out each drive one at a time? 3 cycles of preclear would be a week. Then a day to rebuild disk 3. Then add disks and copy data off disks 1,2,4 (assuming it could be done at same time, that's another day). Is there anything I can do to reduce my Risk level earlier? Quote Link to comment
itimpi Posted January 31, 2016 Share Posted January 31, 2016 A Red x means a write failed and the way to recover from that is to rebuild the disk (not run xfs_repair). Although reallocated sectors are not intrinsically bad it is very bad if the number is not stable and you imply that it is not. Also on new disks I would expect the number to be 0 (or something very close to it). Were all 4 disks showing problems acquired at the same time - maybe you got a batch of bad disks? Normally I would say that reallocated sectors are not caused by external factors, but since you mention they all share the same power adapter I am not sure if that could contribute in any way. I would think that these disks are candidates for a RMA. Are any of the disks showing Pending sectors? If any are then that will mean that a rebuild of any other disk is unlikely to be completely successful. Quote Link to comment
Matt_G Posted January 31, 2016 Share Posted January 31, 2016 All are on a molex to 4 SATA adapter - the other disks 5,6,7,Parity all zero errors on a different adapter. Am I reading this correctly? You have all 4 disks being powered off of a single molex that is being split into 4 SATA power adapters? If so, I'm betting that is the cause of the issue. But I'm probably misinterpreting your post. Quote Link to comment
methanoid Posted January 31, 2016 Author Share Posted January 31, 2016 Something LIKE this http://www.ebay.co.uk/itm/281910580199 Yes... Please tell me how else you get 12 devices powered from a PSU that has 6 SATA plugs? Its a Silverstone 550W Gold PSU. I guess I could make up new modular cables but you still end up with 4 or more HDDs on a single cable run.... There are precious few PSUs with 6 device cable runs... Quote Link to comment
Matt_G Posted January 31, 2016 Share Posted January 31, 2016 The reason I brought it up is because I have heard of people having issues with power splitters/adapters. I find it hard to believe that all your new disks are at fault here, though anything is possible I'm thinking it could be that adapter causing a power issue Do you have another to try? Or temporarily use a second PSU to power some stuff, just for testing purposes? I'm basically just thinking out loud here... Quote Link to comment
methanoid Posted February 1, 2016 Author Share Posted February 1, 2016 Oh sure. I mention it as I think that is the cause. My questions now are how to move on with minimal data loss Quote Link to comment
JorgeB Posted February 1, 2016 Share Posted February 1, 2016 First you should post the diagnostics, may help to better advise on how to proceed. You essentially have 2 options, but before attempting either replace the molex to sata cable just in case, then: 1-rebuild all disks one at a time 2-try to copy the data from each disk to a new one on a test/trial server or on that server by doing a new config without parity (for this option first take a screenshot of the array and don’t write anything more to it in case you have to revert to original config and try a rebuild) Seeing the diagnostics may help to choose which one I would try first but if one fails you can always try the other. Quote Link to comment
Frank1940 Posted February 1, 2016 Share Posted February 1, 2016 ... All are on a molex to 4 SATA adapter - the other disks 5,6,7,Parity all zero errors on a different adapter. ... Is there anything I can do to reduce my Risk level earlier? ... I would get that Molex to 4 adapter out of the computer and pull hard on each and every wire. I tend to suspect that you have a defective one. You might also inspect the PS molex connector in the same manner. A second thought is the wire gauge that is used in the adapter. Is it at least as large as the wire coming into the PS molex? ( A second line of reasoning, given the cost, is to buy a new one from a different source and pitch the old one.) About reducing risk--- Least risk, turn the server off and don't use until you get the new disks precleared. If that is not possible, don't write to the array. Simple reads shouldn't make the problem worse. Remember with unRAID as long as you can access the disk, you won't lose everything off a defective but readable disk. Quote Link to comment
Frank1940 Posted February 1, 2016 Share Posted February 1, 2016 One other thought. You have an i7 with 32GB of RAM, Video card, and eight-to-twelve hard disks on a 550W PS. Have you thought you might have exceeded the total power rating of that PS and it is doing funny things to protect itself? Quote Link to comment
Bjonness406 Posted February 1, 2016 Share Posted February 1, 2016 Something LIKE this http://www.ebay.co.uk/itm/281910580199 Yes... Please tell me how else you get 12 devices powered from a PSU that has 6 SATA plugs? Its a Silverstone 550W Gold PSU. I guess I could make up new modular cables but you still end up with 4 or more HDDs on a single cable run.... There are precious few PSUs with 6 device cable runs... Buy a new psu Here is a list of psu with 12+ sata port (double check), it is in norwegian, but you can see the list of psu there. http://www.prisjakt.no/category.php?o=produkt_pris_inkmoms&m=s257290637 Quote Link to comment
methanoid Posted February 3, 2016 Author Share Posted February 3, 2016 Thanks for the advice guys One other thought. You have an i7 with 32GB of RAM, Video card, and eight-to-twelve hard disks on a 550W PS. Have you thought you might have exceeded the total power rating of that PS and it is doing funny things to protect itself? No, I had actually swapped the video card out for a HD5450 (19w) and I always double check my theoretical power consumption with online PSU calcs as well as with meters. Not even close to capacity! Buy a new psu Here is a list of psu with 12+ sata port (double check), it is in norwegian, but you can see the list of psu there. http://www.prisjakt.no/category.php?o=produkt_pris_inkmoms&m=s257290637 No, not doing that. If I had 15 or 20 HDDs as many UnRAIDers do... would you then find a 20 SATA power connector PSU and replace a perfectly good PSU? No... ... All are on a molex to 4 SATA adapter - the other disks 5,6,7,Parity all zero errors on a different adapter. ... Is there anything I can do to reduce my Risk level earlier? ... I would get that Molex to 4 adapter out of the computer and pull hard on each and every wire. I tend to suspect that you have a defective one. You might also inspect the PS molex connector in the same manner. A second thought is the wire gauge that is used in the adapter. Is it at least as large as the wire coming into the PS molex? ( A second line of reasoning, given the cost, is to buy a new one from a different source and pitch the old one.) About reducing risk--- Least risk, turn the server off and don't use until you get the new disks precleared. If that is not possible, don't write to the array. Simple reads shouldn't make the problem worse. Remember with unRAID as long as you can access the disk, you won't lose everything off a defective but readable disk. Yup. Tossed it and used another to be sure. Four new HDDs on Preclear now and a 5th on desk to keep as a spare. Quote Link to comment
kizer Posted February 3, 2016 Share Posted February 3, 2016 What make/model of PowerSuppy are you using? Quote Link to comment
methanoid Posted February 3, 2016 Author Share Posted February 3, 2016 What make/model of PowerSuppy are you using? Silverstone Strider Gold 550W http://www.silverstonetek.com/legacy.php?pid=351&area=en&model=ST55F-G%20(V1.X)&tno=2# According to http://outervision.com/power-supply-calculator my settings for PSU Load Wattage: 294 W Recommended UPS rating: 600 VA Recommended PSU Wattage: 344 W Since I intend to uprate on CPU and graphics cards I will be swapping for a Coolermaster Silent Pro 800W I have shortly. 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.