3

Using the sample recipe I am deploying a WordPress solution on EC2:

juju bootstrap --constraints "instance-type=t1.micro"
juju set-constraints instance-type=t1.micro
juju debug-log
juju deploy mysql wordpress-db
juju deploy wordpress
juju add-relation wordpress wordpress-db
juju expose wordpress

After this I point my browser to the public url and get the error:

502 Bad Gateway
nginx/1.1.19

Looking at the debug log I see this error:

2012-11-10 20:08:19,083 unit:wordpress/0: hook.output INFO:  * Restarting PHP5 FastCGI Process Manager php5-fpm
2012-11-10 20:08:20,267 unit:wordpress/0: hook.output INFO:    ...fail!

sshing to the WordPress box, I tried restarting the php service but I get a seg fault:

ubuntu@ip-10-194-31-77:~$ sudo /etc/init.d/php5-fpm start
Segmentation fault (core dumped)

Anybody know the issue? Is there a problem with the php version and WordPress?

1 Answer 1

2

This is probably a problem with RAM. If you look on your wordpress hosts, they're likely running out of RAM quite quickly, and probably spraying 'out of memory' errors in dmesg. Thats because the default tuning setting of single does include a lot of plugins.

If you run juju set wp-service-name tuning=bare it will use less RAM and it might be able to survive on a t1.micro. If that still doesn't work, I'd recommend filing a bug against the charm here:

https://launchpad.net/charms/+source/wordpress/+filebug

And perhaps look into patching it to allow reducing the number of php5-fpm processes running or other things to save RAM in such a hyper-constrained environment.

2
  • Thanks that tuning trick seems to have fixed the issue!!! Nov 11, 2012 at 19:22
  • Tyler, perhaps Accept the answer then, so this doesn't appear as an "unanswered" question?
    – SpamapS
    Nov 13, 2012 at 1:25

You must log in to answer this question.

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