Ad Widget

Collapse

php-fpm 100% CPU

Collapse
X
Collapse
 
  • Page of 1
  • Time
  • Show
Clear All
new posts
Previous template Next
  • weerok
    Junior Member
    • Nov 2013
    • 4
    #1

    php-fpm 100% CPU

    Hello,

    While updating (actually migrate from Centos to FreeBSD), strange bug appears:

    If open Web-> SCREENS->Zabbix server (default screen from install), PHP-FPM loads to 100% some CPU Cores and do not goes down.
    Other pages seams to not load to 100% PHP.

    Code:
    last pid: 36750;  load averages:  2.79,  5.15,  8.80                                                                                       up 0+06:22:03  21:23:37
312 processes: 4 running, 308 sleeping
CPU: 25.8% user,  0.0% nice,  4.4% system,  0.2% interrupt, 69.5% idle
Mem: 4848M Active, 1828M Inact, 1595M Wired, 3310M Buf, 11G Free
Swap: 8192M Total, 8192M Free

  PID USERNAME    THR PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
34562 www           1 103    0   195M 39404K CPU3    3   3:04 100.00% php-fpm: pool www (php-fpm)
34564 www           1 103    0   191M 32840K CPU6    6   3:04 100.00% php-fpm: pool www (php-fpm)
 1290 mysql       184  20    0 15343M  4068M uwait   3  78:24   0.98% [mysqld]
 1757 zabbix        1  20    0   528M   177M nanslp  7   4:44   0.00% zabbix_server: history syncer #17 [synced 4 items in 0.01

    Same nginx/zabbix/php-fpm/my.cnf config was used on new server.
    Same hardware (with VMWare).

    Current Server config:
    Code:
    FreeBSD zabbix-access-bsd9 9.3-RELEASE-p2 FreeBSD 9.3-RELEASE-p2 
zabbix 2.2.7
nginx version: nginx/1.6.2
PHP 5.5.17 (cli) (built: Oct 13 2014 12:37:24) 
Zend Engine v2.5.0, Copyright (c) 1998-2014 Zend Technologies
mysql  Ver 14.14 Distrib 5.1.73
perl -v  This is perl, v5.16.3
    On Centos config, this issue was not observed.
    Code:
    CentOS release 6.4 (Final)
zabbix 2.2.0
nginx version: nginx/1.0.15
PHP 5.3.3 (cli) (built: Jul 12 2013 20:35:47) 
Zend Engine v2.3.0, Copyright (c) 1998-2010 Zend Technologies
mysql  Ver 14.14 Distrib 5.1.66
perl -v      This is perl, v5.10.1

    Please help what can be tuned.
    Tags: None
    • weerok
      Junior Member
      • Nov 2013
      • 4
      #2
      didnt find the answer, but after deleting this screen, and recreate screen with same graphs, problem didnt appear again.

      Suppose some graph had linux dependencies (ethX), and after migration they disappeared, causing to loop infinitly some process.

        Comment

        • tukusejssirs
          Junior Member
          • Feb 2020
          • 1
          #3
          I experience this with Zabbix 5.4.3. There are 9-10 paralel `php-fpm: pool zabbix` process, each consuming about 20-25 % of a CPU core. Zabbix monitors only a single server (the one it is running on) and about 50 graphs each consisting of 4 boolean values. We use CentOS 8.4.2105 (non-Stream) with 6 vCPUs, 30 GiB RAM on a physical server where no other VM is running.

          How can I debug this?

            Comment

            Previous template Next
            Working...