Results 1 to 5 of 5

Thread: I/O Problems with server

  1. #1
    Join Date
    Apr 2006
    Posts
    8
    Rep Power
    9

    Default I/O Problems with server

    Hi

    We have a server with the following specs
    4 CPUS Xeon at 3.0GHz
    2G RAM
    250G SATA disc

    Zimbra 4.5.4
    RedHat 4
    accounts: 430

    The problem we are currently having is the following:
    top - 11:44:19 up 1:30, 1 user, load average: 9.54, 9.70, 9.70
    Tasks: 253 total, 1 running, 252 sleeping, 0 stopped, 0 zombie
    Cpu(s): 3.1% us, 1.8% sy, 0.0% ni, 61.0% id, 34.0% wa, 0.1% hi, 0.0% si

    Load Average is high as well as the wait time which means the IO is really slow right now.

    avg-cpu: %user %nice %sys %iowait %idle
    8.88 0.00 2.34 32.02 56.76

    The queue of course is high:

    [zimbra@correo ~]$ qshape active | head
    T 5 10 20 40 80 160 320 640 1280 1280+
    TOTAL 815 209 236 370 0 0 0 0 0 0 0

    We already turned off the logger, but it did not help much.

    Is there anything we could do to improve the IO. Or at least find out exactly why is so high?

    Thanks

  2. #2
    Join Date
    Jan 2007
    Location
    Rochester, NY
    Posts
    45
    Rep Power
    8

    Default

    The first thing I see that pops up at me is that / (root) /swap and /opt are all on the same disc?

    Split them up. If you start swapping at all then it's going to wreck your IO performance. hanving / on a seperate drive is just good practice as well.

    4 3Gig Xeons and you only have 2GB of ram as well? Spend a few bucks and toss some more memory in as well while you are at it.

    I will be a couple of (brief) outages if you can swap ram and disk quickly. Add another drive.. boot the machine without starting ZCS, copy over /opt to the new disk, change the mounts, expand / (or put /var in the freed space, I don't see how you have your partitioning set up, so i cant really say). Reboot. (I'm oversimplyfying.. but thats how I would handle this)
    Last edited by AimanA; 04-13-2007 at 10:26 AM.
    It only takes a few flakes working together to unleash an avalanche of destruction.
    chown -R us ./yourbase
    http://www.aimana.net

  3. #3
    Join Date
    Apr 2006
    Location
    Williamsburg, VA
    Posts
    451
    Rep Power
    9

    Default

    I am by no means a performance guru, the two things that stands out to me and AimanA points out or alludes to both of them:

    RAM: It is cheap..put more in there.

    Disc: You are only running a single disc in there? Lots of discs will improve your IO. Zimbra recommends RAID 10 for installs over 100 users. How many discs can your chassis take? I would do what AimanA suggests, but in a slightly larger scale. Max out the discs in your chassis, raid them together in a 10 and then drop the box, don't start zimbra, setup the new mount point, copy the current /opt to the new disc array you just added and then reshuffle the mount points and bring up zimbra from the new array under /opt

  4. #4
    Join Date
    Jan 2007
    Location
    Rochester, NY
    Posts
    45
    Rep Power
    8

    Default

    I neglected to mention.... if neither of the two above solutions help your problems, you can try tripping the 'noatime' flag in your fstab on any Ext or Reiser filesystems.

    If you don't know what I mean in the above statement, and what the consequences are, don't do it.
    It only takes a few flakes working together to unleash an avalanche of destruction.
    chown -R us ./yourbase
    http://www.aimana.net

  5. #5
    Join Date
    Apr 2006
    Posts
    8
    Rep Power
    9

    Default RE: I/O Problems with server

    Hi all

    Thanks for the advice

    Indeed all the instalation was on a single harddisk (swap, tmp, zimbra, root). This is because we installed as it is from a DataCenter (the planet).

    I tried to follow AimanA advice... copied zimbra to another harddisk and then mounted it as /opt but for some weird reason zimbra refused to start... It just get stuck starting MTA... heck I could not even re-installl. Ill try again later.

    Meanwhile what I did do was move other partitions to the other disk (Swap, tmp, backups) and today it behaved better.

    Maybe we should switch to a better server with scsi disks... and more ram.

    Thanks for the help, while we get a better server ill try to tweak as much as possible.

Similar Threads

  1. initializing ldap...FAILED(256)ERROR
    By manjunath in forum Installation
    Replies: 39
    Last Post: 06-07-2013, 10:27 AM
  2. Zimbra fails after working for 2 weeks
    By Linsys in forum Administrators
    Replies: 10
    Last Post: 10-07-2008, 12:42 AM
  3. need advice on configuring zimbra to work with fax server
    By pheonix1t in forum Administrators
    Replies: 0
    Last Post: 07-11-2007, 07:46 PM
  4. Ubuntu 6.06 LTS server install problems
    By Kavey1978 in forum Installation
    Replies: 17
    Last Post: 05-30-2007, 01:49 AM
  5. Error 256 on Installation
    By RuinExplorer in forum Installation
    Replies: 5
    Last Post: 10-19-2006, 09:19 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
  •