January 8, 201115 yr After the power got tripped in my house (faulty surge protector) I ran my parity check as usual (run one every month) only to find that I've got 2,700 errors and counting (about 15% of the drive still to do). Some quick searching and reading on the forums tell me that these errors will be automatically fixed - i.e. it's not just "you've got errors", it's "I found and fixed some errors". Just a parity drive and a single data drive thus far in my setup. Is that correct? One of those silly newbie questions that I promised myself to try and compile as a wiki page for the other newbies out there...
January 10, 201115 yr Author Hate /bumping my own thread (feels like something a teenager would do ) but I'm still wary of this problem. I've run 2 parity checks now (with a 3rd ongoing). Each reports back that there are no errors, but the error count next to the parity drive keeps increasing - 13,000 and counting. Feels a little confusing, so I'm doubting what I'm reading: unRaid has found and fixed these errors for me, and I need to stop panicing... ...and yes, I'm now in the market for a UPS - most likely a APC Power-Saving ES 550VA. Great way to learn how important a UPS is....
January 10, 201115 yr Hate /bumping my own thread (feels like something a teenager would do ) but I'm still wary of this problem. I've run 2 parity checks now (with a 3rd ongoing). Each reports back that there are no errors, but the error count next to the parity drive keeps increasing - 13,000 and counting. Feels a little confusing, so I'm doubting what I'm reading: unRaid has found and fixed these errors for me, and I need to stop panicing... ...and yes, I'm now in the market for a UPS - most likely a APC Power-Saving ES 550VA. Great way to learn how important a UPS is.... OK, there are two different error types involved. One is an error in reading a disk. It is shown on the line for the parity drive on the management web-page. It will increment every time a sector on the affiliated disk is reported by the disk as unreadable. It sounds as if this is the counter that continues to increment every time you perform a parity check. The other error count is for parity errors. It is displayed on the bottom third of the screen where it typically says something like: Last checked on 1/1/2011 2:16:26 PM, finding 0 errors. If the errors are "read" errors, affiliated with the parity disk, odds are you are having un-readable sectors on that disk and unRAID is doing its best to deal with them by re-writing the sectors and hoping the SMART firmware on the disk will use sectors from its spare pool of sectors. Most large disks have several thousand spare sectors, but if the number keeps growing, basically your parity disk is slowly crashing. To verify run a SMART report on the parity disk. You 'll need to do this on the raw device. ( the /dev/sd? device affiliated with the parity drive) The command will be smartctl -d ata -a /dev/sdX where sdX = the device for your parity disk. (You can see the three letter device name on the "devices" page. It will be the three letter device in parens looking something like this: : (sda) " Look for sectors pending re-allocation, or re-allocated sectors. The count will be in the "raw" column on the far right on the respective lines. Post the whole smart report if you do not know what to look for. Oh yes, if I'm right, now would be a really good time to order a replacement drive. Joe L.
January 10, 201115 yr Author Edit: Incredibly rude not to thank a guru for dropping by - so thank you! Well the good news is that the SMART report said "Passed" - however the screenshot doesn't inspire confidence: I seemed to remember seeing those Pre-fail messages when I ran preclear on the initial build (I even posted one of the results for comment). So a bit more wiki investigating and I ran the short smartctl test: smartctl -d ata -tshort /dev/sda Which gave me this: If I understand this correctly (and I REALLY want confirmation or mocking as appropriate - it's the only way to learn! ) I've got 2 reallocated sectors but a ton of read errors. I'm wondering therefore if the SMART firmware has done it's job and was able to work around it. The bulk of the read errors probably came from my first parity check following the power failure (the UI flagged up around 11,000) which might explain the high number of read errors. This plus the fact that no errors have been logged is giving me a little hope... I also did the same thing on the other drive in my setup (I'm planning to add drives as required) and got no reallocated sectors or read errors. I'm thinking the longer SMART test might be worth running, but I await the advice of a guru before proceeding!
January 10, 201115 yr You have 1449 sectors pending re-allocation the next time they are written. Those 1449 sectors are unreadable. The type column is a category of the parameter. Old age types are expected to incriment over the life of the drive, pre-fail are those that might give an indication of an impending failure. What is interesting is that your drive apparently is not changing the "VALUE" very much from its initial value (probably 200) on the pending-reallocation column. To me, that drive has already failed. As I said earlier, time to order a new drive. The good news, it is your parity drive and your data drives are probably just fine. Joe L.
January 10, 201115 yr Those pending sectors could be a result of the power outage and they may be written back to their original locations. The pending count can go down without increasing the reallocated count. This is just a more optimistic theory and Joe is probably right.
January 10, 201115 yr Those pending sectors could be a result of the power outage and they may be written back to their original locations. The pending count can go down without increasing the reallocated count. This is just a more optimistic theory and Joe is probably right. Unlikely but possible. You can stop the array, un-assign the parity drive, start the array with it un-assiged, stop it once more, re-assign it, and let it get completely re-written. It will just write what is already there, except for those sectors pending a "write" to learn if they can be written back to their original locations. Then you'll know for sure. Joe L.
January 10, 201115 yr Author I'll give it a try first - if nothing else it will help my learning curve. Two thoughts though: is it as literal as you've made it sound in case I get stuck part way through? Secondly, is there any danger to the existing data? If there's an element of risk then I might as well grab a new HD.
January 10, 201115 yr Currently you are essentially without parity protection. As long as another drive does not fail your data is safe. You need to rebuild parity in any case. I would try to rebuild on your existing drive. If the pending sectors do not go away or become reallocated then you should order a new drive. If the sectors become reallocated your drive may still last from months to years but to be safest you should replace it. Have a drive pre-cleared and ready to install. I use my spare drive as a cache drive...
January 10, 201115 yr I'll give it a try first - if nothing else it will help my learning curve. Two thoughts though: is it as literal as you've made it sound in case I get stuck part way through? Secondly, is there any danger to the existing data? If there's an element of risk then I might as well grab a new HD. If another disk were to fail you can force the parity disk back into use. Remember, we are just re-writing what is already there... You would be no worse off than you are right now. Joe l.
January 12, 201115 yr Author Did all of that: - After remounting the parity drive I ended up with no errors (good start!) - Then I re-ran a parity check and picked up 150 or so errors on the UI (uh-oh....) - smartctl (short test) shows up 1265 reallocated sectors (doh!) I really don't need messages like: "SMART overall-health self-assessment test result: FAILED! Drive failure expected in less than 24 hours. SAVE ALL DATA." ...and I really don't need to post screen shots either to confirm things! So guess I'm shopping on-line in a day or two! I was quietly hoping that it would be salvagable as a data drive, but I seriously doubt that's an option either...
Archived
This topic is now archived and is closed to further replies.