1

I'm running an AWS marketplace bitnami image for wordpress. The site has been running for a few months and recently started crashing. Something's leaking memory and I can't find out what it is.

I've updated php-fpm settings

pm = ondemand
pm.max_children = 20
pm.process_idle_timeout = 10s
pm.max_requests = 500

And I've drop max_memory in php.in to 64M

But I can still see the memory slowly disappearing.

bitnami@ip-172-31-7-85:/opt/bitnami/php/etc$ free -m
             total       used       free     shared    buffers     cached
Mem:          2000       1236        764         75         94        489
-/+ buffers/cache:        652       1348

The httpd and php-fpm processes just take more and more memory till the site falls over.

bitnami@ip-172-31-7-85:/opt/bitnami/apache2/logs$ ps aux --sort -rss
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
mysql     1478  0.6 11.0 1392736 227120 ?      Sl   19:35   0:24 /opt/bitnami/mysql/bin/mysqld.bin --defaults-file=/opt/bitnami/mysql/my.cnf --basedir=/opt/bitnami/mysql --datadir=/opt/bitnami/m
daemon    3841  1.1  3.0 1305952 62988 ?       Sl   20:36   0:01 /opt/bitnami/apache2/bin/httpd.bin -f /opt/bitnami/apache2/conf/httpd.conf -DDISABLE_BANNER
daemon    4045  4.3  2.5 286040 51928 ?        S    20:36   0:06 php-fpm: pool wordpress                                                                                                          
daemon    3843  0.7  2.4 1305496 50564 ?       Sl   20:36   0:01 /opt/bitnami/apache2/bin/httpd.bin -f /opt/bitnami/apache2/conf/httpd.conf -DDISABLE_BANNER
daemon    4047  3.5  2.4 286012 50356 ?        S    20:36   0:05 php-fpm: pool wordpress                                                                                                          
daemon    3842  0.8  2.4 1305892 50020 ?       Sl   20:36   0:01 /opt/bitnami/apache2/bin/httpd.bin -f /opt/bitnami/apache2/conf/httpd.conf -DDISABLE_BANNER
daemon    4048  0.6  2.3 1297712 48532 ?       Sl   20:36   0:01 /opt/bitnami/apache2/bin/httpd.bin -f /opt/bitnami/apache2/conf/httpd.conf -DDISABLE_BANNER
daemon    4118  3.0  2.1 285992 44916 ?        R    20:36   0:04 php-fpm: pool wordpress                                                                                                          
daemon    4116  3.9  2.1 286000 44856 ?        S    20:36   0:06 php-fpm: pool wordpress                                                                                                          
daemon    4115  1.4  1.6 273832 34420 ?        S    20:36   0:02 php-fpm: pool wordpress                                                                                                          
root      3831  0.0  1.1 206004 23300 ?        Ss   20:36   0:00 /opt/bitnami/apache2/bin/httpd.bin -f /opt/bitnami/apache2/conf/httpd.conf -DDISABLE_BANNER
root      3793  0.0  0.3 267808  7836 ?        Rs   20:35   0:00 php-fpm: master process (/opt/bitnami/php/etc/php-fpm.conf)

The sites under very low load. But it's only taking 10 mins before it crashes.

I'm getting the following in my error_log

[Thu Mar 09 18:43:49.420787 2017] [pagespeed:warn] [pid 1549:tid 140430532409088] [mod_pagespeed 1.9.32.14-0 @1549] Fetch timed out: http://ourdomaincom/wp-content/uploads/2015/01/ourdomain_page_3.jpg-200x200.jpg (connecting to:172.31.7.85) (1) waiting for 50 ms
[Thu Mar 09 18:43:49.583085 2017] [pagespeed:warn] [pid 2201:tid 140430112970496] [mod_pagespeed 1.9.32.14-0 @2201] Fetch timed out: http://ourdomain.com/wp-content/plugins/LayerSlider/static/js/layerslider.transitions.js?ver=5.6.9 (connecting to:172.31.7.85) (1) waiting for 50 ms
[Thu Mar 09 18:43:49.584180 2017] [pagespeed:warn] [pid 1548:tid 140430381405952] [mod_pagespeed 1.9.32.14-0 @1548] Fetch timed out: http://ourdomain.com/wp-content/themes/thefox/js/custom_woo_js.js?ver=4.6.1 (connecting to:172.31.7.85) (1) waiting for 50 ms

How can I fix this or track down the memory leak?

Improve this question
3
  • Those errors are generated by Pagespeed, you could try turning that off. Why would fetching static resources time out? Is it sending requests for static content to PHP? You haven't included configuration so we can't check. You could use dynamic PHP, or configure PHP to restart the worker process after X invocations (pm.max_requests). You could switch from Apache to Nginx, Nginx is very lean, takes 15MB RAM (plus 175 virtual) on my server.
    – Tim
    Commented Mar 9, 2017 at 21:09
  • Thanks @Tim - That's helpful. I'll try turning off mod_pagespeed. Also I'll look at config - it does seem that php is trying to serve static content.
    – Aidan Ewen
    Commented Mar 9, 2017 at 21:12
  • It is very unlikely to be a memory leak - far more likely that the system is just overloaded - mod_pagespeed and wordpress on a box with with 2Gb of memory? You should also be checking the nfconntrack limits.
    – symcbean
    Commented Mar 11, 2017 at 21:56

1 Answer 1

Reset to default
1

Turns out the site had been hacked.

I migrated it into the hands of the fine folks at WPEngine. They've been amazing, identifying the problem, and sorting it quickly.

It's not been a good introduction to wordpress. Kind of hope I never have to look after another wp installation :(

Improve this answer
1
  • 3
    wordpress: a remote code execution API that comes with blogging functionality as a free extra.
    – Rob Moir
    Commented Mar 11, 2017 at 19:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .