Backuppc and Load Average

Discussion in 'Server Operation' started by Tripple2, Oct 3, 2007.

  1. Tripple2

    Tripple2 New Member

    I'm using Backuppc for quite a while but there's one issue I was never been able to solve. A very high load average.

    top - 22:08:57 up 34 days, 2:38, 1 user, load average: 4.42, 4.45, 4.39
    Tasks: 110 total, 7 running, 102 sleeping, 0 stopped, 1 zombie
    Cpu(s): 89.0% us, 8.3% sy, 0.0% ni, 2.7% id, 0.0% wa, 0.0% hi, 0.0% si
    Mem: 5076616k total, 5020284k used, 56332k free, 148k buffers
    Swap: 4088500k total, 124k used, 4088376k free, 3024432k cached

    I added a screenshot. As you can see, after # /etc/init.d/backuppc reload, the Cpu drops to 0% for about 1 hour. After that, back to my high numbers.

    The fileserver/backupserver is a Dell Poweredge 2950, Mandriva Powerpack + 2006 with 1GB RAM. I added 4GB so now I'm having 5GB. I was hoping adding more ram would solve this but all remains the same.

    Any ideas where to start looking for a sollution?
     

    Attached Files:

  2. falko

    falko Super Moderator ISPConfig Developer

    Can you see (in the top output) which processes are consuming the most performance?
     
  3. Tripple2

    Tripple2 New Member

    My server is going crazy. I made a cronjob every 30 min:
    /etc/init.d/backuppc reload > /dev/null

    This is the result from top:
    top - 16:35:05 up 34 days, 21:04, 1 user, load average: 151.07, 150.45, 148.37
    Tasks: 265 total, 2 running, 263 sleeping, 0 stopped, 0 zombie
    Cpu(s): 1.7% us, 2.7% sy, 0.0% ni, 95.7% id, 0.0% wa, 0.0% hi, 0.0% si

    I removed the cronjob and reboot the server.
     

    Attached Files:

  4. Tripple2

    Tripple2 New Member

    I'm not sure backuppc is the cause. I see a lot of these:

    27747 root 23 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    30522 root 22 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    863 root 20 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    3683 root 22 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    6568 root 24 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    9355 root 22 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    12147 root 23 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    14925 root 22 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    17717 root 24 0 2740 436 356 D 0 0.0 0:00.00 smartctl
    20503 root 24 0 2740 436 356 D 0 0.0 0:00.00 smartctl
    23292 root 24 0 2740 436 356 D 0 0.0 0:00.00 smartctl
    28356 root 22 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    31098 root 24 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    1353 root 23 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    4055 root 24 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    6871 root 23 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    9570 root 23 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    12243 root 22 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    14920 root 22 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    17709 root 23 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    20482 root 24 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    23263 root 24 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    26017 root 23 0 2740 432 356 D 0 0.0 0:00.00 smartctl
    28800 root 23 0 2736 428 356 D 0 0.0 0:00.00 smartctl
    31587 root 24 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    1918 root 22 0 2736 432 356 D 0 0.0 0:00.00 smartctl
    7736 root 23 0 2740 432 356 D 0 0.0 0:00.00 smartctl
     
  5. falko

    falko Super Moderator ISPConfig Developer

    Can you try to switch off smartctl?
     
  6. Tripple2

    Tripple2 New Member

    Stopping smartd failed. I reboot the server without smartd. Now I have to wait for a while to see what will happen. I'll post the result here.
     
  7. Tripple2

    Tripple2 New Member

    Stopping smartd got my server back under control. But there's still a high load average and cpu%.

    top - 20:10:44 up 3:38, 1 user, load average: 4.52, 4.40, 4.36
    Tasks: 111 total, 5 running, 105 sleeping, 0 stopped, 1 zombie
    Cpu(s): 91.2% us, 7.7% sy, 0.0% ni, 1.2% id, 0.0% wa, 0.0% hi, 0.0% si
    Mem: 5076616k total, 358748k used, 4717868k free, 760k buffers
    Swap: 4088500k total, 0k used, 4088500k free, 189040k cached

    When I stop Backuppc, cpu drops to 0%, load average 0.03
    No idea why smartd suddenly gone crazy but backuppc must be the cause.

    All I want is to take a nightly backup from 6 servers. 4 remote, 2 in the same office as my backup server.

    I think backuppc_dump must be the real cause. Since I only want to take a backup at night, why is there 24/24 a process called backuppc_dump.
     
  8. falko

    falko Super Moderator ISPConfig Developer

    Maybe you should start BackupPC only when you need it (e.g. via cron).
     
  9. Tripple2

    Tripple2 New Member

    I've been thinking about that but this is only a workaround. I want to be able to access backuppc during the day to check logfiles, restore files, etc.
     

Share This Page