First request always fails

[ripatte Profile]

Hello, we're using Zoo to run a RoR application in an EC2 instance.

For some reason, when our application has been on idle or if application hasn't been accesses since deployment, the very first request always fails and user gets an HTTP/500 error.

I've already specified worker request timeout:

Code: Select all

<add name="WORKER_REQUEST_TIMEOUT" value="1000" />

We use ruby.1.9.pipe transport:

Code: Select all

<add name="rails.project#x64" scriptProcessor="ruby.1.9.pipe"  path="*" verb="*" modules="HeliconZoo_x64" preCondition="bitness64" resourceType="Unspecified" requireAccess="Script" />

We use IIS to serve static files in production mode:

Code: Select all

<remove name="rails.project#x64" />

I'm wondering if we're doing something fundamental wrong here.
Any ideas how to fix this?

[Yaroslav]

It would be nice to see what is covered under this 500 error. Would you enable DEPLOY_LOG and ERROR_LOG_DIR if your current 500 error page does not provide specific error information.
Also it does seem like EC2 instance is set up to terminate worker process on idle. Ruby applications are known to require long start time and AWS administration system may drop connection if response was not received in some specific delay.

[ripatte Profile]

Hello and thanks for your reply.

These two entries occur in the log files:

ZooError.log:

Code: Select all

[6/26/2012 20:45:12]   ZooApplication I/O Error:
JobFastCgi:[/] _requestId:7, BACKEND_READ_COMPLETE, POST_MODE_PARTIAL


production.log:

Code: Select all

Connecting to database specified by database.yml
  SQL (0.0ms)  USE [vision]


[Yaroslav]

Seems like your error and production logs has been truncated bu forum engine. Please provide complete logs to the [email protected]

[gavind Profile]

+1. I'm looking for the same answer.

[Yaroslav]

Then please provide us with error log. Sometimes you can read more information if you get full error message from Helicon Zoo and select 'expand all fields'.