SSD

There were some issues under which the non-correcting parity check would have come in very handy. Imagine running a parity check that generates thousands and thousands of parity errors. Might indicate that a disk was corrupted, and perhaps should be rebuilt based on parity rather than vice versa. In such a case you would not want to blindly update parity. I also remember a user that got a few parity errors every time he ran a check. Didn't know if it was parity or a disk that was causing the issue (memory tests were good). If it was disk failing, updating parity was actually reducing the ability to recover. Being able to perform non-correcting checks would allow diagnostics to continue without creating more corruption. I suggested that parity check should always be non-correcting, but should remember the location of each parity anomaly. If everything looked good and there were a couple of parity errors that the user could rationalize, he/she could request that all be corrected (with the locations known, this would be near instant). But if the parity errors were extensive, the user would have some tools to know what file was on each disk at that parity error location to try to isolate the issue. This was quite a while ago, but this is what I am remembering.

Smart report says the airflow temperature has failed in the past. Not a good thing. Do you have adequate drive cooling?

Did you mean 1.12beta and 1.13 will correctly preclear a 3TB drive?

I have used 1.12 to preclear 5 3TB drives and have had no problems. Is it safe to assume that even if 1.12 was not correct, that unRAID properly formatted the drives anyway? Not sure - but it may have just been fixing this ... http://lime-technology.com/forum/index.php?topic=4068.msg140029#msg140029

In addition, if you are having what appear to be disk related errors, you should also include a smart report for each drive giving errors. See instructions at link below: http://lime-technology.com/wiki/index.php?title=Troubleshooting#Obtaining_a_SMART_report

Modify the original post in a thread to change the displayed title.

No worries. I was able to add the drive to my array. Have a great vacation! Seems like Irene is cooperating!

Joe L. - I precleared a 3T drive with 1.12 beta a while back. Tonight I was confirming the disk was indeed precleared before adding it to my array. I downloaded the latest and greatest preclear (1.12), and tested the preclear signature, and it said the disk was NOT precleared. I then went back to version 1.12beta and it says the disk IS precleared. Is this just a little bug, or was there a change in the preclear signature? If so, what version was the change made? Is the disk precleared? Thanks for your help! Here is the preclear -t output from the 1.12beta Pre-Clear unRAID Disk /dev/sdn ################################################################## 1.12 Device Model: Hitachi HDS5C3030ALA630 Serial Number: MJ1323YNG0Zxxx Firmware Version: MEAOA580 User Capacity: 3,000,592,982,016 bytes Disk /dev/sdn: 3000.6 GB, 3000592982016 bytes 255 heads, 63 sectors/track, 364801 cylinders, total 5860533168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000 Device Boot Start End Blocks Id System /dev/sdn1 1 4294967295 2147483647+ 0 Empty Partition 1 does not end on cylinder boundary. ######################################################################## ========================================================================1.12 == == DISK /dev/sdn IS PRECLEARED with a GPT Protective MBR == ============================================================================ Here is the same output with version 1.12 Pre-Clear unRAID Disk /dev/sdn ################################################################## 1.12 Device Model: Hitachi HDS5C3030ALA630 Serial Number: MJ1323YNG0Zxxx Firmware Version: MEAOA580 User Capacity: 3,000,592,982,016 bytes Disk /dev/sdn: 3000.6 GB, 3000592982016 bytes 255 heads, 63 sectors/track, 364801 cylinders, total 5860533168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000 Device Boot Start End Blocks Id System /dev/sdn1 1 4294967295 2147483647+ 0 Empty Partition 1 does not end on cylinder boundary. ######################################################################## ========================================================================1.12 == == Disk /dev/sdn is NOT precleared == 1 4294967295 5860533104 ============================================================================

2. I disabled the interrupt inside the card's firmware Why ? The interrupt is just used to be able to boot from drives on the card, and tends to mess up boot order to the USB.

No "more" command is included in the distro.

Me too! Thanks madburg! Flawlessly simple. Two small tips ... 1. Instead of removing the flash at step 31, I was able to enter the command ... type adapters.txt (before powering down) and quickly hit the pause key (might take few tries to catch it, but I was lucky and got it first time). There was the SAS address, which I wrote down. Lazily saved me a trip upstairs to read the USB on another computer. 2. I disabled the interrupt inside the card's firmware

Does this really remove the -d ATA option which unMENU adds to each smartctl command? It is this which causes problems with many add-on controllers. Yes Not sure if it used to be required but is not now, or if it was just the convention. As stated above, I provided a way to override the default behavior in myMain. Not sure about unmenu itself.

In myMain, you can add the custom drive attribute "smartopt" and set it to "-A" on each drive connected to the controller. Let me know if you have trouble.

Nice build! If you are getting enough airflow through the drives without needing the 60mm fans then you are doing EXTREMELY well! But to get that airflw you must have sealed up the case vents pretty effectively, because air will take the path of least resistance, and pulling air through the drive cages is normally NOT the least resistance! Maybe skipping the trays lakes it an easier "pull" than with other cages. Still don't understand why they would use dual 60mm fans instead of one 92mm fan? Dual 60mm are probably more expensive, louder, and move much less air. As for the comment of logorithmic hearing - a loud fan is a loud fan. I once installed a 40mm fan on my chipset. So proud of myself for the clever mounting technique. Computer was a little loud, but I had several large fans and figured some noise was the cost of good cooling. 6 months later the 40mm fan was screaming like a banshee and I had to pull it. When I pulled it out, I couldn't believe how quiet the computer was. All along that 40mm fan was creating a lot of extra noice - right square in the midde of the case. I can promise that 6 60mm fans (3 cages), if you used them, would be far from silent!

6 pending sectors may not be a problem, but could indicate that the drive is doing to die an early death. I would run the preclear for another cycle on that drive. I would expect the 6 to be reallocated and no new ones to be detected. If every preclear cycle / parity check increases the number of bad sectors, I'd RMA the drvie. The other drives look ok. I have also responded in your general support posting related to the other errors you are seeing.

Notice that the "value" and "worst" columns both have the value of 100. That means, from the manufacturer's perspective, 0 and 65536 are equally healthy. In binary, 65536 is 10000000000000000. So one bit is turned on. That bit may have a special significance to the manufacturer that means nothing to us. Your disk is good. Use it and enjoy your array!

There is a poll somewhere that asks about whether people are finding drive issues on 1st, 2nd, or 3rd preclear pass. Most are found in 1st round. Remember any drive can fail at any time - preclears are no guarantee the drive won't fail soon after, we just think it makes it less likely. Anyway, I run 1 pass.

Sounds like an issue with your unRaid server and not with your drive. If the drive loses power, that has been known to crash a server. Other crashes can occur if the syslog gets too large.

Try Notepad2 a lightweight text editor. Does not even need to be installed.

No disk needs to be precleared. Preclear does a pretty decent burn in test of any drive. In the end, assuming the preclear works properly, the drive is left with a preclear signature. The preclear signature is ONLY useful if you are adding the disk as an array disk. Having the preclear signature prevents unRAID itself from clearing the disk (a somewhat painful process becuase your array is offline during the lengthy clearing process). So no, it is not necessary to preclear a cache disk, but you should do it anyway. All of its contents will be lost (if it had any), but you will have confidence that the disk is functioning within normal parameters before you start using it.

I'd preclear this guy two more times. It is very unusual for a drive with more than a handrul of reallocated sectors to not continue to have more sectors reallocated with most every preclear or parity check cycle. If you can go three times in a row, I'd be satisfied that the drive is solid.

You can determine the state of the array with this command ... /root/mdcmd status|strings|grep "mdState"

As I had thought, the post-read verify failed. What preclear does is first read the entire disk, then zero the entire disk, and then read the entire disk verifying that it is full of zeros. What happened is that last step found some location(s) where the data was not zero. Could have litterally been 1 bit in 20 Trillion - that's all it takes. As Joe L. says, there are numerous things that can cause this. I'd recommend running an overnight memory test as a starting point. If you have bad or misconfigured memory, it can cause problems during the write or the post-read phase. Could also be a bad or loose data cable or some incompatibility. These can be hard to find, but not always. Do the memory test and based on the results we can suggest additional tests to try to narrow it down.

myMain echoes the status returned from preclear. My guess is there was a post-read verification error, but will have to look at the actual preclear report to be sure.

Love it but would never work in my house. Kids, animals, wife, me - someone would touch the wrong thing and kablam!