wbee Posted April 30, 2015 Share Posted April 30, 2015 Hi. Building a new unRAID server and after some read errors from a WD green i replaced the disk with a 2TB Samsung 5400rpm 2,5". The complete array only have 1 movie in it, 1,3Gb otherwise it is free from files. But the rebuild estimated finish is like 14 days!! I thought my 8 year old Drobo was slow with 70 hours to rebuild. As you can se in picture the CPU is an AMD BE 2350 with 2,1Ghz and dual core and it is running at 100%. And if i try to watch the movie through smb it stutters and lagging ofcourse. Seems strange that the unRAID server would be useless for up to 14 days while rebuilding. Is is supposed to be this slow to rebuild? Or will it be faster, or am i missing something? Thanks for your support. Link to comment
itimpi Posted April 30, 2015 Share Posted April 30, 2015 There is something not working properly as rebuild speeds should be more like 100MBps. It is typically a drive acting up. You need to provide at least a syslog to have any chance of helping with diagnostics. SMART reports on your drives would be a good idea as well. Link to comment
JonathanM Posted April 30, 2015 Share Posted April 30, 2015 Need a syslog to even guess accurately, but my WAG is that you have a second drive with issues. Unraid needs all other array drives to be read perfectly to reconstruct 1 failed drive, so if any of your other drives are having an issue, it will slow things to a crawl. Also, drive reconstruction uses the most power of almost anything you can do with the tower, so the PSU could be slightly underpowered and causing drive read errors. Did the drive you replaced have bad sectors? Perhaps the drive wasn't really bad, it could have been cables, power, or something else. Link to comment
wbee Posted April 30, 2015 Author Share Posted April 30, 2015 The replaced WD Green drive had read errors and failed the Smart tests. I also had an used ssd drive as cache that stopped working. All other disks was ok on smart test. Before I putted it all together again i ran Wdidle3 and disabled "intellidrive" feature on the 2 WD Green that i have in array. The good part is that i do not have any data to loose in array, the bad part about this error is that i really want to sort this out so i can trust unRAID. I can ofcourse clear all data and start over the parity but then it feels like i give up and do not learn anything about this. Syslog attached and it says something about an bus error that i don't understand. https://drive.google.com/file/d/0B_LxjLam19RCUmx5bmFNdTJkWlE/view?usp=sharing Link to comment
itimpi Posted April 30, 2015 Share Posted April 30, 2015 That syslog is saying that the drive SAMSUNG HD103UJ, S13PJ1KPC03628 is misbehaving. It appears to have dropped offline although that could just be a cabling issue. Link to comment
Squid Posted April 30, 2015 Share Posted April 30, 2015 Apr 30 11:54:16 server kernel: ata6.00: cmd 25/00:00:60:2b:00/00:02:00:00:00/e0 tag 0 dma 262144 in Apr 30 11:54:16 server kernel: res 51/84:7f:e1:2b:00/84:01:00:00:00/e0 Emask 0x10 (ATA bus error) Apr 30 11:54:16 server kernel: ata6.00: status: { DRDY ERR } This error is in your log, along with a bunch of READ DMA EXT. It sounds like when you replaced a drive you may have inadvertently knocked a cable loose, and your 1TB samsung is having connection problems. It's still working, but has downgraded itself speed wise to be PIO instead of DMA. I've never seen the READ MULTIPLE EXT errors that are popping up, but I have a feeling that its because the drive is now in PIO mode. I would check the power and sata connections to the drive and at the mobo. Link to comment
wbee Posted April 30, 2015 Author Share Posted April 30, 2015 Yes, it did missbehave but i thought i fixed that. At first the computer did not rekognice that disk. I shutted down the server and checked cables to it and reconnected them. Then it worked and i started the rebuild. If it still is problem like you say then i might need to replace the sata cable. I will try to replace it and report back. Link to comment
RobJ Posted April 30, 2015 Share Posted April 30, 2015 The ICRC error flag almost always indicates a bad SATA cable (the other possibility is bad power to the drive). Throw that cable away! Dealing with all of the error timeouts, and the drive having been slowed to the lowest PIO level, makes the process incredibly slow, and 13 days may be optimistic! Link to comment
wbee Posted April 30, 2015 Author Share Posted April 30, 2015 Now the array is good and feels healthy again. Impressed by your help and knowledge! Thanks for keeping my trust up in this. I will now start to transfer data to the array as soon as the rebuild is done. Changed the sata cable and that fixed the problem. Now cpu is pendling normal between 65-98% and i can also stream a movie from array without stuttering at the same time. Rebuild is around 100Mb/s. This is more like it :-) https://drive.google.com/file/d/0B_LxjLam19RCRUdzaEZQQ0hHT2M/view?usp=sharing Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.