lovswr Posted February 12, 2013 Share Posted February 12, 2013 Went from 5.0-rc5 to 5.0-rc11 & put new 3TB drive which will replace the existing 2TB parity drive. I have a case with a "hot" SATA port so I put the new drive there so I can start preclearing. Only when I do ./preclear_disk.sh -l the new drive does not show up. It is listed & connected as /dev/sde. Has something changed in regards to preclearing? Is there some kind of way to do this from the web GUI now? Link to comment
Frank1940 Posted February 12, 2013 Share Posted February 12, 2013 Current version 1.13. Are you using it? Link to comment
PeterB Posted February 12, 2013 Share Posted February 12, 2013 You seem to be of the impression that the preclear script is part of unRAID. It isn't - it's provided by a third party, not supplied with unRAID, and will never change just because you upgrade the unRAID version. Link to comment
kamhighway Posted February 12, 2013 Share Posted February 12, 2013 It is listed & connected as /dev/sde There may be other ways to do it but I think you have to say which drive you want to preclear like this: cd /boot preclear_disk.sh /dev/sde Link to comment
lovswr Posted February 13, 2013 Author Share Posted February 13, 2013 You seem to be of the impression that the preclear script is part of unRAID. It isn't - it's provided by a third party, not supplied with unRAID, and will never change just because you upgrade the unRAID version. Yes that was my impression. Read up on the preclear configuration tutorial & my 3TB is now underway. Also had a flaky 2TB SAMSUNG just fail in my array outright. I will get another 2 TB, preclear that drive & replace my failed SAMSUNG before I upgrade the parity to 3TB. Thanks to all who answered. Link to comment
Gog Posted February 14, 2013 Share Posted February 14, 2013 Also had a flaky 2TB SAMSUNG just fail in my array outright. I will get another 2 TB, preclear that drive & replace my failed SAMSUNG before I upgrade the parity to 3TB. Thanks to all who answered. I might be lost here, but isn't a preclear a waste of time on a disk replace? Shouldn't the rebuild process do that anyway? You might want to do it anyway to do a burn test of the drive, but process-wise I was under the impression that it was not necessary. Gog Link to comment
JonathanM Posted February 14, 2013 Share Posted February 14, 2013 I might be lost here, but isn't a preclear a waste of time on a disk replace? Shouldn't the rebuild process do that anyway? You might want to do it anyway to do a burn test of the drive, but process-wise I was under the impression that it was not necessary. Gog You are right, it's not necessary. It's just highly advisable if you are rebuilding on a brand new drive, because of the bathtub curve. I definitely wouldn't call it a waste of time. Link to comment
Gog Posted February 14, 2013 Share Posted February 14, 2013 I'll grant you the "not a waste of time" but I'm so damn nervous when I have a dead drive that I probably would not want to wait the 24+ hours to preclear. I've never had a DOA drive either. I bet I'll learn from that lack of experience eventually... Link to comment
Frank1940 Posted February 14, 2013 Share Posted February 14, 2013 I'll grant you the "not a waste of time" but I'm so damn nervous when I have a dead drive that I probably would not want to wait the 24+ hours to preclear. I've never had a DOA drive either. I bet I'll learn from that lack of experience eventually... The DOA drives are not the worst ones to experience. It's the one that worked when you first put it into service and then fails between two hours and (approximately) hundred hours of being in the array. (Years ago, many computer manufacturers advertised that they 'burned' in all new computers for three days before shipping them to the customer. It was their hope that this burn-in would catch all those early failures! Experience indicated that three days was enough to uncover a very high percentage of the early failures and the computer was now on the flat part of the failure curve where it would run for years without problems until it reached the end-of-life point.) Link to comment
JonathanM Posted February 14, 2013 Share Posted February 14, 2013 I'll grant you the "not a waste of time" but I'm so damn nervous when I have a dead drive that I probably would not want to wait the 24+ hours to preclear. I've never had a DOA drive either. I bet I'll learn from that lack of experience eventually... It would benefit your peace of mind greatly to invest in an extra drive the same size as your parity drive and preclear it multiple times in advance of any failures. Having a cold/warm/hot spare is standard for any RAID array in a typical business setting. It's even better if you have a spare drive bay to leave the drive mounted, powered, and manually spun down. That way when a drive fails, you don't even have to power cycle the server (which can cause other failures at the worst time) to start a rebuild. Link to comment
Joe L. Posted February 16, 2013 Share Posted February 16, 2013 Went from 5.0-rc5 to 5.0-rc11 & put new 3TB drive which will replace the existing 2TB parity drive. I have a case with a "hot" SATA port so I put the new drive there so I can start preclearing. Only when I do ./preclear_disk.sh -l the new drive does not show up. It is listed & connected as /dev/sde. Has something changed in regards to preclearing? Is there some kind of way to do this from the web GUI now? If the drive is assigned to the array, it will not show up when you type preclear_disk.sh -l The actual device names are assigned as the devices initialize and are recognized by the linux OS. A give drive may be /dev/sde one boot, and /dev/sdd the next. This is very common when drives are identical models and initial spin-up times are nearly identical. As far as a new "defective" drive. From posts on this forum, somewhere between 1 out of 10 to 1 out of 5 are defective. If nothing else, run a "long" test on the new drive. smartctl -t long /dev/sdX it will take 4 or 5 hours to complete. It will identify unreadable sectors. Do not assign the drive to the array... (or its spin-down timer will likely abort the test) After the test is complete you can get the results by typing: smartctl -a /dev/sdX unfortunately, it will terminate on the first unreadable sector. If the report shows an un-readable LBA, then a preclear might resolve it. You can look in the system log to learn the actual device name for any given boot up. Joe L. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.