Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

First something about the infrastructure:

We have migrated one of our backend/admin servers to amazon. The applications on this server are only used sporadically (max. 10 parallel users). We are using a t1.micro instance running Gentoo. As Webserver we are using nginx in combination with PHP-FPM. The problem is that sometimes the server is dead slow that means a page needs 10-20sec to load (max 2sec on my local dev environment, and these pages are heavy ... 80mb+). I tried to analyze and reproduce the error but only figured out the memory is okay (270mb free) while cpu load is at 100%


nginx has 2 workers and max. 1024 connections.

PHP-FPM (I removed the comments)

error_log = /var/log/php-fpm.log


; Pool Definitions ; 


listen =

user = www
group = www

pm = dynamic

pm.max_children = 8

pm.start_servers = 2

pm.min_spare_servers = 2

pm.max_spare_servers = 4

pm.max_requests = 100

I hope some one has a clue what to do :)

share|improve this question
up vote 1 down vote accepted

Take a look at where you will find detailed information on t1.micro and how it comares to other instance types (like m1.small). It will help to understand the instance's behaviour.

Short: As soon as you exceed the "short burst" phase you will be getting a high amount of cpu 'st' (shown with top) which is slowing down everything.

share|improve this answer
We changed the instance type now everything works fine :) – Frido Dec 14 '11 at 16:45

EC2 micro instances provide short bursts of CPU, not constant CPU power. This is the expected behavior. micro instances will only provide full CPU power for a few minutes at most then they limit it back to about 5-10% capacity. If you look in top you'll see a high value for 'st', this is CPU steal from the hypervisor (I.E. CPU you can't use/is thin provisioned).

Instances of this family provide a small amount of consistent CPU resources and allow you to burst CPU capacity when additional cycles are available. They are well suited for lower throughput applications and web sites that consume significant compute cycles periodically


share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.