Elandril

Member
  • Content Count

    4
  • Joined

  • Last visited

Community Reputation

0 Neutral

About Elandril

  • Rank
    Member
  1. Elandril

    Silent filesystem corruption

    Sorry for the follow-up post, but I don't seem to be able to edit my previous post... More information about this data corruption bug can be found here.
  2. Elandril

    Silent filesystem corruption

    Do you by any chance use Kernel 2.6.19 ? There has been a serious flaw concerning filesystems in Kernel 2.6.19 which was fixed by Linus Torvalds, the master himself , in 2.6.20 !
  3. Elandril

    Areca 1280 benchmarks

    Wow, that looks like an awesome performer!
  4. I think it is important to clarify what WD means when they are talking about RAID which becomes clear, when you read their TLER specs. Let's look at some situations when a bad sector is encountered: without TLER:non-RAID controller: the drive detects the error and tries to recover it (this can last up to 1 or 2 minutes if the drive is otherwise busy, but usually works within a few seconds). While the drive tries to read this sector the issuing process in the computer "hangs" until the recovery is sucessful/fails (-> error is log) or a timeout period is exceeded (I think about a minute in XP). RAID controller: the drive detects the error and tries to recover it (like above this can take a while). After about 8 seconds the RAID controller patience times out and it identifies this drive as "defective". This can lead to different results depending on the type of RAID. E.g. for a RAID-0 this is fatal and the array is unusable; a RAID-1 or -5 gets put into degraded mode and loses it's redundancy and performance. Even if the drive can recover the error later the drive will still be marked as faulty and a complete rebuild of the array is necessary! [*]with TLER: non-RAID controller: the drive detects the error and tries to recover it, but if the error couldn't be fixed within 7 seconds the drive reports an error back to the controller. Contrary to popular belief the drive doesn't abandon the recovery process after those 7 seconds, but remembers the error and waits for further recovery instructions. Since a normal controller doesn't know anything about error recovery the drive after some time will try again to recover and remap the bad sector (in background). Meanwhile when the controller got the error message the OS logs a disc error and/or notifies the user. The issuing process hangs for a maximum of 7 seconds and the operation fails, if the repair wasn't possible within those 7 seconds. RAID-controller: again the drive detects the error and starts its recovery. If this isn't completed within 7 seconds the RAID-controller get an error notification and the drive waits for further recovery instructions. The controller can now start its own recovery methods: for a RAID-1 this is a simple task of reading the data from the other drive; for RAID-5 this involves recalculating the data from the parity information. Meanwhile the RAID-controller can start a conversation with the drive about the status of the error and they can work on a solution together. This may sound strange, but believe me, modern RAID-controller and drives have very sophisticated methods of recovery. In case of a redundant array (e.g. RAID-1/-5) the controller can tell the drive what data it expects there, so that the drive can swiftly remap the sector and write the appropriate data to the replacement sector. In this case the user doesn't notice any performance/data losses (other than a message in the controller's log) and the drive remains an operating array member. If the array was a RAID-0 however the error causes the array to be faulty and unusable. So the conclusion would be that indeed if you're using a RAID controller you're far better off with TLER, but ONLY if you utilize a redundant array (RAID-1/5/6/...). If you use a RAID-0/JBOD/... you'll lose data anyway - TLER or not. For using a controller it doesn't make much difference since the process that issued the read will most likely fail anyways - if the error couldn't be fixed within 7 seconds, it's very unlikely to be fixed at all. I hope this helps to clear up some of the confusion! I'd recommend this drive in both cases! The 5 year warranty and the 24hr factory stress test are in my opinion a far better deal than the remote chance that an bad sector can be recovered if it wasn't within 7 seconds! Especialy since this drive utilizes a fully bit CRC!