Ad Widget

Collapse

GRAPHS not working anymore

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • Xeno
    Junior Member
    • Jan 2009
    • 15

    #1

    GRAPHS not working anymore

    Hi all,
    i've got a strange problem with the graphs. Expecially with Switches graphs. Everything worked fine but, suddenly graphs do not work anymore.
    If i choose the %CPU graph this is what i've got nothing.

    While if i show the 1 Day graph i see this


    I've seen a very long queue lately




    and the a TOP on the running zabbix server shows:

    7785 mysql 20 0 148m 41m 5628 S 165% <---- %CPU ????? 8.2 55:01.73 mysqld

    and the CPU is used with the 88% almost all the time.

    Everything was working fine when the the day before yesterday i've found zabbix DB Down and the Server had gone down too. I could not started the server because of a DB error in two tables: trends_uint and history_uint. i've dropped them and re-created.
    Any relation???


    PLEASE HELP!!!
    Last edited by Xeno; 06-03-2009, 19:49.
  • nelsonab
    Senior Member
    Zabbix Certified SpecialistZabbix Certified Professional
    • Sep 2006
    • 1233

    #2
    How's your processor load?

    Look into tuning your DB. My first thought is your MySQL queries are timing out.
    RHCE, author of zbxapi
    Ansible, the missing piece (Zabconf 2017): https://www.youtube.com/watch?v=R5T9NidjjDE
    Zabbix and SNMP on Linux (Zabconf 2015): https://www.youtube.com/watch?v=98PEHpLFVHM

    Comment

    • Xeno
      Junior Member
      • Jan 2009
      • 15

      #3
      As i said before:
      TOP on the running zabbix server shows:

      7785 mysql 20 0 148m 41m 5628 S 165% <---- %CPU ????? 8.2 55:01.73 mysqld

      and the CPU is used with the 88% almost all the time.

      PROCESSOR's load is very high!!!

      I've increased inno_db_buffer_pool_size to 768Mb but nothing changed. How can i verify that mysql queries are timing out?
      Any other tip?

      Thanks

      Comment

      • nelsonab
        Senior Member
        Zabbix Certified SpecialistZabbix Certified Professional
        • Sep 2006
        • 1233

        #4
        Turn on log slow queries and set the slow query timer to 2-5 seconds. I don't know the exact conf entries off hand so you'll need to look them up.

        Also search around on the forum there are some good posts on DB tuning along with conf files others have used.
        RHCE, author of zbxapi
        Ansible, the missing piece (Zabconf 2017): https://www.youtube.com/watch?v=R5T9NidjjDE
        Zabbix and SNMP on Linux (Zabconf 2015): https://www.youtube.com/watch?v=98PEHpLFVHM

        Comment

        • Xeno
          Junior Member
          • Jan 2009
          • 15

          #5
          Originally posted by nelsonab
          Turn on log slow queries and set the slow query timer to 2-5 seconds. I don't know the exact conf entries off hand so you'll need to look them up.

          Also search around on the forum there are some good posts on DB tuning along with conf files others have used.
          Thank you very much! i'll post results here!

          Comment

          • Xeno
            Junior Member
            • Jan 2009
            • 15

            #6
            Here we are, i've found a lot of these lines in the log:

            # Query_time: 6 Lock_time: 0 Rows_sent: 2 Rows_examined: 685253
            select value from history_uint where itemid=26274 order by clock desc limit 2;
            # User@Host: zabbix[zabbix] @ localhost []
            # Query_time: 6 Lock_time: 0 Rows_sent: 2 Rows_examined: 685255
            select value from history_uint where itemid=25556 order by clock desc limit 2;
            # User@Host: zabbix[zabbix] @ localhost []
            # Query_time: 6 Lock_time: 0 Rows_sent: 2 Rows_examined: 685252
            select value from history_uint where itemid=25298 order by clock desc limit 2;

            Seams that the hostory_uint table had got some problems; i've deleted everything and let's see what we get now!
            Processor LOAD in now between 12 and 20% very good!!!
            Thanks

            Comment

            • nelsonab
              Senior Member
              Zabbix Certified SpecialistZabbix Certified Professional
              • Sep 2006
              • 1233

              #7
              Cool, good to hear. As you get more data that table will grow, the history tables can grow very fast depending on how much data you're putting into the DB.

              Just curios under Reports -> Status Of Zabbix, what is the Required server performance?
              RHCE, author of zbxapi
              Ansible, the missing piece (Zabconf 2017): https://www.youtube.com/watch?v=R5T9NidjjDE
              Zabbix and SNMP on Linux (Zabconf 2015): https://www.youtube.com/watch?v=98PEHpLFVHM

              Comment

              • Xeno
                Junior Member
                • Jan 2009
                • 15

                #8
                ZABBIX server is running Yes -
                Number of hosts (monitored/not monitored/templates) 68 24 / 0 / 44
                Number of items (monitored/disabled/not supported) 3586 3573 / 8 / 5
                Number of triggers (enabled/disabled)[true/unknown/false] 1455 1449 / 6 [1 / 1 / 1447]
                Number of users (online) 2 1
                Required server performance, new values per second 25.7337 -

                The DB grew up too much once again....hmm i can't delete the history_uint every day...i really need a DB tuning. I'll try tomorrow with a couple of scripts.

                Bye

                Comment

                • Xeno
                  Junior Member
                  • Jan 2009
                  • 15

                  #9
                  After a couple of hours of tunings and checks the LOAD of the CPU on my Zabbix server is finally between 5% and 14%
                  Required server performance is now 14.5
                  I dropped the table history_uint, recreated (i re-created the index too),
                  ran ANALYZE and OPTIMIZE on that table.
                  In the my.cnf i added : innodb_file_per_table
                  and ran mysqlcheck -Aoqp
                  Restarte everything and now Zabbix really works fine
                  I think that the really good tweak was the activation of innodb_file_per_table.

                  Thanks for everything.
                  I hope that this post will be useful for someone else.

                  Comment

                  • nelsonab
                    Senior Member
                    Zabbix Certified SpecialistZabbix Certified Professional
                    • Sep 2006
                    • 1233

                    #10
                    yeah! Glad to know everything worked out. Interesting observation on the innodb option. I wonder if you could use that option to then break out the tables to be on different physical media.
                    RHCE, author of zbxapi
                    Ansible, the missing piece (Zabconf 2017): https://www.youtube.com/watch?v=R5T9NidjjDE
                    Zabbix and SNMP on Linux (Zabconf 2015): https://www.youtube.com/watch?v=98PEHpLFVHM

                    Comment

                    • stiricherai
                      Junior Member
                      • Dec 2008
                      • 4

                      #11
                      Zabbix Graph are empty

                      Dear,

                      Quite few of my graphs are empty. The polling is not as per frequency.

                      Number of hosts (monitored/not monitored/templates) 100 51 / 5 / 44
                      Number of items (monitored/disabled/not supported) 6195 3314 / 28 / 2853
                      Number of triggers (enabled/disabled)[true/unknown/false] 1923 1823 / 100 [19 / 1301 / 503]
                      Number of users (online) 7 1
                      Required server performance, new values per second 80.8134 -

                      I did following already
                      1. my.cnf i added : innodb_file_per_table
                      2. ran mysqlcheck -Aoqp
                      3. Increased some of the frequency of polling for some of the items

                      My box is resonably big -- 4 CPU , 16 GB

                      Still the polling is missing large amout of data. Any help will be highly appericiated

                      Please note
                      CPU Utilization of Zabbix Server < 20 to 25%
                      Memory Utilization of Zabbix server < 20 to 25%

                      Thanks and Regards
                      Sarang
                      Last edited by stiricherai; 19-03-2009, 17:11.

                      Comment

                      • Xeno
                        Junior Member
                        • Jan 2009
                        • 15

                        #12
                        Have you already tried to tune your mysql DB? In this forum you can find a couple of scripts that provide DB tuning

                        Comment

                        Working...