October 24, 20169 yr One of my drives was disabled today. Looking at the log, I'm not sure what this might be. GUI reports "only" 2 read errors. I'm running 6.2.1 and can't update to 6.2.2 right now due to ongoing preclearing of new disk. Can anyone take a look at the log and perhaps provide some input? As this is happening while preclearing a new Toshiba X300 6TB drive, I am wondering if my Sea Sonic G 360 is not providing enough amps and the drive drops out. There's also some - perhaps - unrelated system messages about "program smartctl is using a deprecated SCSI ioctl, please convert it to SG_IO" Have attached log excerpt and overview of drives in system. Thanks in advance for any insights! log.txt tower-diagnostics-20161025-0735.zip
October 25, 20169 yr Community Expert Syslog snippets are seldom sufficient. Go to Tools - Diagnostics and post the complete diagnostics zip.
October 25, 20169 yr Community Expert unRAID will only disable a disk when a write to it fails: Oct 24 12:50:38 Tower kernel: md: disk4 write error, sector=2320118744 SMART for the disabled disk Oct 23 14:57:02 Tower kernel: md: import disk4: (sdj) WDC_WD30EZRX-00D8PB0_WD-WMC4N0E3TKX9 size: 2930266532 does not appear in the diagnostics. Is this disk on a motherboard port? unRAID is currently emulating the disk using parity and all the other disks. All reads and all writes to the disk are actually being calculated from all the other array disks, and the actual disabled disk is not being used. Even if you haven't written anything else to the disk, the write that failed means the disk is out-of-sync. The failed write and any subsequent writes to the disk are in the array because parity was updated, but they are not on the disk. So the disk contents are invalid now. To enable the disk again, you must rebuild it. If we could see the disk in the diagnostics, then we could decide whether to rebuild to the same disk, or to replace it. Check connections, power and SATA, then post another diagnostic.
October 25, 20169 yr Author Yup, I realize that it is being emulated right now. The disk is on a motherboard port. As soon as the preclear of the 6TB finishes, I will be able to check. Will post when I have more. Thanks! Sent from my iPhone using Tapatalk
October 25, 20169 yr Author Done with preclearing - here is a new diagnostics pack. The disk shows up again and SMART test is passed. Wondering if it might be a dodgy SATA cable. Haven't started the array apart from in Maintenance mode to be able to do a reiserfsck check. tower-diagnostics-20161025-2332.zip
October 25, 20169 yr Community Expert SMART looks OK. You can try to rebuild to the same disk. Or replace with the newly precleared 6TB if you want. You should get the array stable one way or another before trying to add a disk to a new slot.
October 26, 20169 yr Author Weird. Disk is mounted, reiserfsck came back clean - but it stays disabled and I don't get the option to rebuild. Wondering if I should try reformatting it and then perhaps I will get the option to rebuild? New diagnostics file attached. tower-diagnostics-20161027-0105.zip
October 27, 20169 yr Community Expert Weird. Disk is mounted, reiserfsck came back clean - but it stays disabled and I don't get the option to rebuild. Wondering if I should try reformatting it and then perhaps I will get the option to rebuild? New diagnostics file attached. Format is NEVER part of a rebuild. If you format the only thing it will be able to rebuild is a formatted disk. Format means "write an empty filesystem to this disk". That is what it has always meant in every operating system you have ever used. unRAID treats this write just like any other, by updating parity. So after the format you will be able to rebuild an empty filesystem, not what you want! Stop array, unassign disk, start array. This will make unRAID forget about the disk assignment. Then stop array, reassign disk, start array to rebuild.
October 27, 20169 yr Author Ok, so what I meant was stopping the array and formatting the disk, then starting it again to provoke the system into recognizing the disk as a replacement. Anyway, I did as you described. Unfortunately, the rebuild was cancelled early as there was now an error reported on another disk (disk2):-/ When I stopped the array, both disk2 and disk4 were reported as missing.. Rebooted server, and they are now present again (with disk4 being reported as disabled - disk2 seems alright). I did swap the SATA cable on the 3TB disk (disk4) that is acting up. Seems kind of unlikely that two SATA cables would fail at the same time, so I will investigate if disk2 and disk4 are on the same controller or sharing the same cable or y-splitter from the PSU to understand if there could be something else at play here. This problem came on the back of swapping out disk2 earlier in the week (http://lime-technology.com/forum/index.php?topic=52996.msg509460#msg509460), which also suddenly was reported disabled. Could these disk problems over the last couple of days be symptoms of a PSU that is starting to fail? Until a week ago, the server was motoring happily along without issues. New diagnostics attached. tower-smart-20161027-0929.zip
November 19, 20169 yr Author Thought I'd just take the time to wrap this up. I have replaced the PSU and that seemed to bring stability back and stop the problems with disks going missing. Disk 4 (3TB) has been replaced with a 4TB disk and rebuilt. Afterwards, I have identified that some files have gone missing on disk2. Disk2 was replaced just before all of this started, so I really can't say if the files were actually missing already because of that. But I have been able to restore them from backup. Array has passed parity check and order seems to be restored I have since introduced a couple of 6TB drives and preparing for an additional parity drive. Thank you to trurl for helping out!
Archived
This topic is now archived and is closed to further replies.