What happens in case of URE when recovering ?

If you are concerned, be sure to purchase Enterprise disks.

http://www.techrepublic.com/blog/datace ... -ures/1752

to answer the question, it really depends on how the disk handles the ure and possibly whether its a read or write error.

worst case scenario, the disk freezes and causes a multiple disk failure and the loss of data, alternatively it could just hang the rebuild process causing a hard reboot to be done. At which point the rebuild process will resume, either repeating the freeze or possibly getting through the error, or an error being generated and the device going into tech support mode.

best case scenario, the error is on a write and the disk automagically reallocates to a new sector and continues on with the rebuild. These will typically be seen as 're-allocated sector count' errors in the smart diagnostics and error logs.

TeknoJnky wrote:

to answer the question, it really depends on how the disk handles the ure and possibly whether its a read or write error. worst case scenario, the disk freezes and causes a multiple disk failure and the loss of data, alternatively it could just hang the rebuild process causing a hard reboot to be done. At which point the rebuild process will resume, either repeating the freeze or possibly getting through the error, or an error being generated and the device going into tech support mode. best case scenario, the error is on a write and the disk automagically reallocates to a new sector and continues on with the rebuild. These will typically be seen as 're-allocated sector count' errors in the smart diagnostics and error logs.

I believe an URE is an "unrecoverable read error" On the SMART stats this will result an increase of the "Current Pending Sector" count. I'm not sure how the NAS will handle this on a resync. However, if it did block the resync, one could clone the drive to a working drive, and then resync with the clone. That would limit the errors to the unreadable sectors. However, figuring out which files were impacted by the URE would be difficult.

Yes, URE is an Unrecoverable Read Error, which means one bit of data irremediably lost. I'm not sure how they're related to the pending sector count, which may or may not be recovered, or remapped to other sectors. I'd like to know specifically for the ReadyNAS NV+ implementation of X-RAID, does this result in a failure of the whole array, or only a stripe ? (I understand that determining which files are affected is not easy, and that's the job of the filesystem anyway).

MrPhelps wrote:

Yes, URE is an Unrecoverable Read Error, which means one bit of data irremediably lost. I'm not sure how they're related to the pending sector count, which may or may not be recovered, or remapped to other sectors. I'd like to know specifically for the ReadyNAS NV+ implementation of X-RAID, does this result in a failure of the whole array, or only a stripe ? (I understand that determining which files are affected is not easy, and that's the job of the filesystem anyway).

Sectors are not reallocated ("remapped") on a read error, that is only done when a write fails. When a read fails, the pending sector count is incremented.

More to your point, there is an old post on the subject:

yoh-dah wrote:

snowman136 wrote: Sooo ... if I understand you, a RAID5 or x-RAID array can rebuild the array after a drive completely fails AND there are some bad spots (unrecoverabe read errors) on the reamining drives. Not exactly. If there's an unrecoverable error on the 2nd disk, the ReadyNAS will go into 'life-support" mode and prevent any more I/O to the data volume. If for some reason the 2nd disk refuses to join in the RAID volume, it's often possible to do so manually, but this requires some expertise and is not automatic.

I am not sure if that information is still correct, or if it applies to all platforms (this quote is from 2007). Hopefully one of the jedi will comment.

Thanks. The ReadyNAS NV+ has not seen an update in a while, so it probably applies. I'd like to know what exactly needs to be done to force a drive back in the array :).

MrPhelps wrote:
I'd like to know what exactly needs to be done to force a drive back in the array :).

Most likely, the only answer you will get is to open an online ticket and let support do it, I doubt very much they are going to release any details that would allow the general public to get in and do potentially destructive commands that could put your data in further jeporady.