ec2 deployment spikes CPU to 100% forever

[njs]

I am trying to do a deployment to a ec2 micro instance. It gets to the
point where it triggers the assets precompile task and after that the
cpu usage on the ec2 goes upto 100% and remains there. Most of the cpu
is used between syslogd and node (javascript engine) processes. This
is a Rails 3.1 app and I installed node from source code on ec2
instance. This cpu usage continues forever until I kill the processes.
I am using capistrano ver. 2.9.0.


What could be going on here?

-Navjeet

[Lee Hambley]

I am trying to do a deployment to a ec2 micro instance.

Could be a bundler problem, take a look at the bundler notes/docs/list for the discussions about using it in low-memory situations, otherwise …

 

It gets to the point where it triggers the assets precompile task and after that the

cpu usage on the ec2 goes upto 100%.

Your node process is making horrible use of the memory. 

Sounds like a node issue to me, just because Cap is calling the command doesn't mean we can help.

[Jim Morris]

Are you using rails 3.1.0? Because 3.1.1 fixed a bug where the rake asset:precompile was broken when run on some servers. The bug did not seem to have anything to do with capistrano.

[Lee Hambley]

Jim, good point, I didn't even think about that!

[njs]

It is Rails 3.1.1. Once I switched from Node.js to SpiderMonkey as
javascript engine things are fine. So I am assuming it is Node
related.