Application hanging in production! New Relic Phy Mem Utilization Haywire! Help!

Rails
1

Hi all,

My app was doing fairly well in production past couple a months. But
past 2 days, its gone off its rocker! Not able to figure out what the
problem is. The symptoms are as follows…

  • It first went down when I deployed and ran a migration yesterday. A
    simple column addition to a table
  • The production log showed the SignalException::SIGTERM error ( dont
    know anything about it )
  • a server reboot seems to solve the problem temporarily
  • New Relic RPM shows wayward physical mem utilization.
  • im on Slicehost and the ‘free’ command shows way too much SWAP
    memory usage (which is basically not good)

Does it sound like im having a memory leak? anywhere i could look?
anything i can do??
Oh those who have fought the battle and come out victorious… bestow
upon me ur knowledge! :smiley:
Seriously… any pointers?

Thanks!

2

Hey blasterpal,
already on it!
more information in case u can help. I run a cron on my server which
basically accesses a particular URL in my Rails app which performs
some tasks. it seems to take loads of time… and when i drill down in
New Relic, the following line is thrown up…

Dependencies.new_constants_in(Object) { super }

this is a line from active_support

File: /Library/Ruby/Gems/1.8/gems/activesupport-2.1.0/lib/
active_support/dependencies.rb

i saw a couple of error throws caused due to dependency loading from
active_support in my production log earlier yesterday during initial
analysis.

Any more ideas?

  • Hide quoted text -

On Thu, Jun 11, 2009 at 5:26 PM, blasterpal [email protected] wrote:

Ram,

I’ve gotten burned recently with production issues not related to
yours. But I only want to add the following. Rollback your changes and
see if the problem disappears for a few days. If so, then you know
where to start looking: at your new code and how it interacts with the
database, Rails ,etc.

H