FedoraForum.org - Fedora Support Forums and Community
Results 1 to 4 of 4
  1. #1
    Join Date
    Jul 2010
    Location
    western PA
    Posts
    111

    raid-check sucks!

    Personalities : [raid6] [raid5] [raid4]
    md1 : active raid5 sdd1[0] sdc1[4] sdg1[2] sdh1[1]
    2930280960 blocks super 1.2 level 5, 512k chunk, algorithm 2 [4/4] [UUUU]
    [================>....] check = 84.0% (821075504/976760320) finish=38.4min speed=67536K/sec


    It's really lame how much raid-check bogs down my system. I mean the thing has 8 cores, 16 HTs, and 24G of RAM. There's really no good excuse for my X session events to take 10-20 seconds to process a simple button click on a GUI while the raid-check is going on in the background. So as expected, two of the 16HTs are pegged and disk state is at 100% IO wait. That should NOT affect X11 response or processing of mouse/keyboard events, but it does.

  2. #2
    Join Date
    Feb 2005
    Location
    London, UK
    Posts
    644

    Re: raid-check sucks!

    Quote Originally Posted by tempest766
    So as expected, two of the 16HTs are pegged and disk state is at 100% IO wait. That should NOT affect X11 response or processing of mouse/keyboard events, but it does.
    Why shouldn't it? If any resource is waiting for IO then why would other resources not also have issues trying to get IO? This is more an issue with architecture rather than processing power, if there's an accident on the road then buses will be effected just as much as cars, if one bus is blocking a road you could have 1 or 100 cars still unable to get to their destination.

    Anyway, you can change the NICE setting in /etc/sysconfig/raid-check to make it, well, nicer.

  3. #3
    Join Date
    Jul 2004
    Posts
    309

    Re: raid-check sucks!

    storage vs probability
    not sure if I believe these numbers

    http://www.enterprisestorageforum.co...e-Numbered.htm

  4. #4
    Join Date
    Jul 2010
    Location
    western PA
    Posts
    111

    Re: raid-check sucks!

    Quote Originally Posted by HaydnH
    Why shouldn't it? If any resource is waiting for IO then why would other resources not also have issues trying to get IO? This is more an issue with architecture rather than processing power,.
    Actually, I'd expect it to depend upon the type of IO. Disk IO channel should not interfere with low bandwidth stuff like keyboard and mouse, so yes...big architectural faux-pa, IMHO.

    Problem is moot now. Four disk stripe now exists where software RAID-5 once lived...and yep. volume is imaged weekly and work is pushed upstream in git daily.

Similar Threads

  1. Weekly raid-check never checks md1
    By llowrey in forum Using Fedora
    Replies: 2
    Last Post: 23rd February 2011, 06:22 PM
  2. Hardware RAID - Check Status?
    By Kilo in forum Using Fedora
    Replies: 5
    Last Post: 20th September 2006, 01:47 AM
  3. RAID Check and Backup
    By il_boss in forum Using Fedora
    Replies: 0
    Last Post: 26th November 2005, 11:47 AM
  4. Howto check RAID is working ok
    By da644 in forum Using Fedora
    Replies: 2
    Last Post: 18th February 2005, 12:20 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •