Ad Widget

Collapse

Slow Frontend Zabbix Performance

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • Kalin35
    Junior Member
    • Sep 2011
    • 3

    #1

    Slow Frontend Zabbix Performance

    We currently just setup a zabbix server with the configuration below. Majority of the time, the web front end takes a while to load pages. Sometimes graphs take a long time to load or it is just a basic page (such as discovered members) that takes a while to load. Any ideas what could be going on?

    Current Configuration:
    2x2.8GHz
    6GB Memory
    Zabbiz 1.8.6
    PostgreSQL 9.0


    PostgresSQL Tunning:
    shared_buffers = 2048MB
    effective_cache_size = 3584MB
    checkpoint_segments = 32
    work_mem = 128MB
    maintenance_work_mem = 32MB

    Zabbix Stats:
    Number of hosts (monitored/not monitored/templates) - 66 / 1 / 42
    Number of items (monitored/disabled/not supported) 5794 / 1109 / 32
    Number of triggers (enabled/disabled)[problem/unknown/ok] 2095 / 20 [4 / 3 / 2088]
    Required server performance, new values per second 233.22 -


    The memory, CPU, etc looks to be fine at all times. I don't see any spikes or anything. Anyone have any other tunning ideas to increase performance?

    Thanks!
  • guesommer
    Junior Member
    • Feb 2009
    • 4

    #2
    Frontend runs PHP, so it might be something in your PHP/apache settings... Thans now just a guess, but your settings look quite fine.

    Also the database might be of intrested. Have you set the memory settings in postgres?

    Br, G.

    Originally posted by Kalin35
    We currently just setup a zabbix server with the configuration below. Majority of the time, the web front end takes a while to load pages. Sometimes graphs take a long time to load or it is just a basic page (such as discovered members) that takes a while to load. Any ideas what could be going on?

    Current Configuration:
    2x2.8GHz
    6GB Memory
    Zabbiz 1.8.6
    PostgreSQL 9.0


    PostgresSQL Tunning:
    shared_buffers = 2048MB
    effective_cache_size = 3584MB
    checkpoint_segments = 32
    work_mem = 128MB
    maintenance_work_mem = 32MB

    Zabbix Stats:
    Number of hosts (monitored/not monitored/templates) - 66 / 1 / 42
    Number of items (monitored/disabled/not supported) 5794 / 1109 / 32
    Number of triggers (enabled/disabled)[problem/unknown/ok] 2095 / 20 [4 / 3 / 2088]
    Required server performance, new values per second 233.22 -


    The memory, CPU, etc looks to be fine at all times. I don't see any spikes or anything. Anyone have any other tunning ideas to increase performance?

    Thanks!

    Comment

    • 0siris
      Member
      Zabbix Certified Specialist
      • Nov 2010
      • 76

      #3
      I'm not into pgsql, but can you post what the pgsql threads are doing while waiting for a response from the gui?
      Also, can you post the output of this command:
      Code:
      vmstat 1 10

      Comment

      • Colttt
        Senior Member
        Zabbix Certified Specialist
        • Mar 2009
        • 878

        #4
        what of kind of harddisk do you use?? and wich filesystem??

        the best at the moment is SSD+ext4
        Debian-User

        Sorry for my bad english

        Comment

        • guesommer
          Junior Member
          • Feb 2009
          • 4

          #5
          Also ensure to use a battery backed write cache on your raid controller, this can improve the write by ten times

          Comment

          • Kalin35
            Junior Member
            • Sep 2011
            • 3

            #6
            Anyone have any ideas of any PHP or Apache settings to look at? All the PHP settings are pretty much what are recommended by Zabbix during the installation.

            Also, I truned on pgsql slow query logs that logs any SQL that takes longer than 200ms and I only saw 1 slow query during the whole entire day.

            Does anyone else have any other ideas? The system resources are barly being touched as well.

            Are there any zabbix_server.conf settings that could cause this? I pretty much have the defaults. I don't see anything being queued either.


            Here is the output of vmstat 1 10 - this was ran while the dashboard was loading which took a little while:

            20:50:50 # vmstat 1 10
            procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------
            r b swpd free buff cache si so bi bo in cs us sy id wa st
            0 0 0 1229812 435004 1915752 0 0 2 14 51 19 0 0 99 0 0
            0 0 0 1229812 435004 1915752 0 0 0 0 1070 200 0 0 99 0 0
            1 0 0 1228076 435004 1915768 0 0 0 12 1086 248 4 0 95 0 0
            0 0 0 1229696 435004 1915768 0 0 0 24 1057 305 6 1 92 0 0
            0 0 0 1229580 435004 1915776 0 0 0 352 1190 480 5 1 94 0 0
            0 0 0 1229580 435008 1915772 0 0 0 188 1064 229 0 0 100 0 0
            0 0 0 1229588 435008 1915840 0 0 0 0 1090 191 0 0 100 0 0
            0 0 0 1229588 435008 1915840 0 0 0 80 1061 171 0 0 100 0 0
            0 0 0 1229596 435008 1915840 0 0 0 0 1087 212 0 0 100 0 0
            0 0 0 1229596 435008 1915840 0 0 0 176 1070 340 2 1 98 0 0

            Comment

            • Colttt
              Senior Member
              Zabbix Certified Specialist
              • Mar 2009
              • 878

              #7
              you can tune you apache2.. remove all you not used modules. and use mod_deflate

              and take a lokk at top when you go to zabbix and look which process is high in top..

              or take a look at google
              Debian-User

              Sorry for my bad english

              Comment

              Working...