Results 1 to 10 of 10
  1. #1
    Join Date
    Jun 2004
    Location
    Delft
    Posts
    99

    New Dual Opteron with Load above 3 in CPanel

    Dear WHT'ers,

    I have a new Dual Opteron Server
    (Opteron 244, 2GB RAM, 2xHD)

    I just started using it and have almost no accounts on it.
    Somehow the load on the server showing up in CPanel Server Status is a red dot with 3.5.

    Here is the output from top.
    18:47:23 up 9 days, 20:37, 1 user, load average: 3.45, 3.18, 2.93
    117 processes: 115 sleeping, 1 running, 0 zombie, 1 stopped
    CPU states: cpu user nice system irq softirq iowait idle
    total 4.0% 0.0% 1.1% 0.0% 0.0% 0.6% 94.3%
    cpu00 1.6% 0.0% 1.0% 0.0% 0.0% 0.4% 97.0%
    cpu01 6.4% 0.0% 1.2% 0.0% 0.0% 0.8% 91.6%
    Mem: 2025028k av, 1811096k used, 213932k free, 0k shrd, 190492k buff
    1403004k actv, 223704k in_d, 32932k in_c
    Swap: 6291432k av, 0k used, 6291432k free 1108620k cached

    PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME CPU COMMAND
    15724 nobody 15 0 15644 15M 4836 S 1.7 0.7 0:00 0 httpd
    13949 nobody 17 0 24628 24M 5092 S 1.0 1.2 0:05 0 httpd
    13950 nobody 15 0 18152 17M 5060 S 0.2 0.8 0:04 0 httpd
    15743 root 23 0 4788 4788 3160 S 0.2 0.2 0:00 1 exim
    15747 root 22 0 4780 4780 3160 S 0.2 0.2 0:00 1 exim
    15714 root 15 0 1304 1304 976 R 0.1 0.0 0:00 1 top
    15737 nobody 15 0 8932 8932 4248 S 0.1 0.4 0:00 0 httpd
    15746 mailnull 23 0 4816 4816 3180 S 0.1 0.2 0:00 1 exim
    1 root 15 0 124 124 52 S 0.0 0.0 0:09 0 init
    ..
    ..
    ..

    I restared Apache, MySQL and FTP thinking maybe it has something to do with it. Still no go. Load is very high..

    Strange thing is there is there are around 20 sites running from it where one is really active and the others are dns only sites.

    Any ideas?
    Thanks in advance.
    http://www.MKEweb.com
    The Netherlands

  2. #2
    Join Date
    Feb 2004
    Location
    Your Screen
    Posts
    3,998
    What happens when you reboot it? Loads low again?

    BTW a load of 3 on a dual Opteron isn't "high," that would be high on a small single processor box like a Celeron or a Sempron, but not on a dual Opteron. You will be able to run a load of 10 or 12 and not feel it, thanks to the piles of brute CPU under the hood. However a load a 3 is higher than obviously you'd like... I see that you've dipped significantly into swap in the past, question is why?

    It would also be helpful to install something like HotSanic http://www.eth0.us/?q=hotsanic which will give you a graphical representation of what is going on in many aspects of your server... particularly I/O, netstat, etc. ... these are things that will drive loads up very quickly.

    Bailey
    Let's Connect on Twitter! @thatsmsgeek2u || Fighting mediocrity one thread at a time.

  3. #3
    Join Date
    Mar 2004
    Location
    New York City
    Posts
    995
    A high load is generally not indicative of a slow/overworked CPU, but rather slow/overworked drives. My first instinct is someone running a poorly optimized MySQL heavy site, or something of the like that's hosing the drives.

    Regards,
    Sam
    Sam Machiz / Director, Product Development / Ubersmith
    smachiz[at]ubersmith.com / [direct] 212-812-4194

  4. #4
    Join Date
    Jun 2004
    Location
    Delft
    Posts
    99
    Thanks for the info peeps.

    bithost(NET) I will take al ook at those programs.

    Somehow I just went out and when I came back the load was normal.

    I read when you had a dual processor server that CPanel and/or linux did not check the load together ... something about having a load of 2 meaning 1 was for each server... But am not sure if this applies.


    I did not want to reboot as it was "rush hour" here. But I will reboot tonight to make sure all the "zombie" processes are gone.
    http://www.MKEweb.com
    The Netherlands

  5. #5
    Join Date
    Feb 2004
    Location
    here and there
    Posts
    746
    Well you can look at your top output right there and see that cpu0 (1) is 97% idle, and cpu1 (2) is 91.6% idle.

    Load is the number of jobs (things in the RUN/RUNNING state at once) in the run que at 5, 10, and 15 minute averages.

    From the lack of CPU utilization, lots of free RAM and no swap usage, I'd say a process/script/something is starting/stopping repeatedly really fast. Login and see, sheesh, does nobody do anything from the console anymore? (Could also be a whole lot of mail being delivered, lots of processes chucking it around, etc)

    Awknet - DDoS Mitigation, Upstream ACLs/Filtering, Unmanaged Dedicated Servers, BGP IP Transit & More!

  6. #6
    Join Date
    Feb 2003
    Location
    Kuala Lumpur, Malaysia
    Posts
    4,974
    Try to find if there is any unusual process running over there. How many sites are you hosting with that? Are those busy sites?

  7. #7
    Join Date
    Feb 2004
    Location
    here and there
    Posts
    746
    Originally posted by UltraUnixNET
    Try to find if there is any unusual process running over there. How many sites are you hosting with that? Are those busy sites?

    You'll have to excuse me for seeming so grumpy today, but, did you read his post? heh...

    Awknet - DDoS Mitigation, Upstream ACLs/Filtering, Unmanaged Dedicated Servers, BGP IP Transit & More!

  8. #8
    Join Date
    Jan 2004
    Location
    North Yorkshire, UK
    Posts
    4,163
    Originally posted by bithost(NET)


    <<< BTW a load of 3 on a dual Opteron isn't "high," that would be high on a small single processor box like a Celeron or a Sempron, but not on a dual Opteron. You will be able to run a load of 10 or 12 and not feel it >>>

    Bailey
    I've managed 30 on an Athlon 64 3000+ without even knowing it had happened, those machines just seem to keep on going ...

    Could possibly be an I/O problem like Sam mentioned, are you using RAID at all? (I've seen something like this with bad RAID drivers before).

  9. #9
    Join Date
    Jun 2004
    Location
    Delft
    Posts
    99
    Oke let me try to answer each question. Could be usefull in the future for others.

    bithost(NET) - No reboot as the server was online just around 4pm dutch time.
    What do you mean with
    "I see that you've dipped significantly into swap in the past, question is why?".
    Thanks for the HotSanic tip. I will look into it.

    jwr- I indeed preffer to use normal shell instead of WHM. That's why I thought maybe it was a misconfiguration in WHM/CPanel. What else could I use instead of top and ps to see what is using resources? (not a linux guru by far but willing to learn!)

    UltraUnixNET- The only processes I saw where httpd, sql, ftp.. usual server stuff. No the server had around 10 sites. Only one being active with a very low visitor count (around 500 per day). Others where just new sites. (Server is new)

    RazorBlue - Dan- I am using Linux Raid 1 with two SATA Maxline III Plus HD's. They are on a Sil3114 Raid Controller. Machine is using Centos 3.5 (2.4 kernel).
    command : vi /proc/mdstats gives

    Personalities : [raid1]
    read_ahead 1024 sectors
    Event: 2
    md0 : active raid1 sdb1[1] sda1[0]
    104320 blocks [2/2] [UU]

    md1 : active raid1 sdb2[1] sda2[0]
    244091520 blocks [2/2] [UU]

    unused devices: <none>
    ~


    After restarting the just an hour ago the load went to average 0.12

    Even thought this solved the problem I most times like to understand what caused it. So if you guys have tips on reading log files, etc all are welcome. Thanks to all btw for the answers !
    http://www.MKEweb.com
    The Netherlands

  10. #10
    Join Date
    Feb 2004
    Location
    here and there
    Posts
    746
    I'm going to stick with mail processing, but, after looking at your 1.0 and 1.7% cpu usage on apache, also say there was a sudden burst in traffic to a site. Setup the server-stats or server-status directive, when you see spikes 'lynx localhost/server-status/' and check, maybe a site is just seeing a lot of traffic.

    'mailq' - see howmuch crap your mailsystem is tossing about at the moment.

    iostat (not sure if this is the same on linux?) - figure out if something is doing a lot of disk access
    systat (not sure if this is the same on linux?) - lots of stuff about a lot of stuff.


    This is the second or third post about CentOS and strange load in the last few days, interesting...

    Awknet - DDoS Mitigation, Upstream ACLs/Filtering, Unmanaged Dedicated Servers, BGP IP Transit & More!

Posting Permissions

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