Huge memory usage after upgrading to 2.4.0 #260

Closed
modsaid opened this Issue May 23, 2012 · 4 comments

Comments

Projects
None yet
2 participants

modsaid commented May 23, 2012

Greetings guys,

I'm not really sure if this is related to the twitter gem. but it is highly probable

After upgrading to the latest 2.4.0 this morning, my passenger workers started using huge memory (4GB+, while normally it used to use 150MB)... after a while of trials and reverting to old code. i noticed that things went back normal after downgrading to 2.3.0 (although i'm not using the gem heavily at all).

I cannot grab the exact problem yet. but I thought of reporting it here, just in case someone faced the same problem:
I'm using rails 3.0.3 on ruby 1.9.2

Here is my complete bundle list

  • abstract (1.0.0)
  • actionmailer (3.0.3)
  • actionpack (3.0.3)
  • activemodel (3.0.3)
  • activerecord (3.0.3)
  • activeresource (3.0.3)
  • activesupport (3.0.3)
  • ancestry (1.2.5)
  • arel (2.0.10)
  • aws-s3 (0.6.2)
  • builder (2.1.2)
  • bundler (1.1.3)
  • cocaine (0.2.1)
  • comma (3.0.3)
  • curb (0.8.0)
  • daemons (1.1.8)
  • domain_name (0.5.3)
  • erubis (2.6.6)
  • escape (0.0.4)
  • eventmachine (0.12.10)
  • factory_girl (2.2.0)
  • factory_girl_rails (1.3.0)
  • faraday (0.8.0)
  • fastercsv (1.5.4)
  • feedzirra (0.0.24)
  • flexible-nested-form (0.0.3)
  • god (0.12.1)
  • i18n (0.6.0)
  • json (1.5.3)
  • kgio (2.7.4)
  • loofah (1.2.0)
  • mail (2.2.19)
  • mechanize (2.5.1)
  • memcache-client (1.8.5)
  • mime-types (1.18)
  • multi_json (1.3.5)
  • multipart-post (1.1.5)
  • mysql2 (0.2.18)
  • net-http-digest_auth (1.2)
  • net-http-persistent (2.6)
  • newrelic_rpm (3.3.2.1)
  • nokogiri (1.5.2)
  • ntlm-http (0.1.1)
  • object_mock (0.1.3)
  • paperclip (2.4.1)
  • polyglot (0.3.3)
  • pr_geohash (1.0.0)
  • rack (1.2.5)
  • rack-mount (0.6.14)
  • rack-test (0.5.7)
  • rails (3.0.3)
  • railties (3.0.3)
  • raindrops (0.8.0)
  • rake (0.8.7)
  • redis (2.2.1)
  • redis-namespace (1.0.3)
  • resque (1.17.1)
  • resque-loner (1.0.1)
  • resque-scheduler (1.9.9)
  • rest-client (1.6.7)
  • rsolr (0.12.1)
  • rufus-scheduler (2.0.16)
  • sax-machine (0.1.0)
  • shoulda (3.0.1)
  • shoulda-context (1.0.0)
  • shoulda-matchers (1.0.0)
  • simple_oauth (0.1.8)
  • sinatra (1.2.8)
  • sitemap_generator (3.1.1)
  • sunspot (1.2.1)
  • sunspot_rails (1.2.1)
  • test-unit (2.4.8)
  • thin (1.3.1)
  • thor (0.14.6)
  • tilt (1.3.3)
  • treetop (1.4.10)
  • twitter (2.3.0)
  • tzinfo (0.3.32)
  • unf (0.0.5)
  • unf_ext (0.0.4)
  • unicorn (4.2.1)
  • vegas (0.1.11)
  • webrobots (0.0.13)
  • will_paginate (3.0.3)
  • xml-simple (1.1.1)
  • yahoo_weatherman_extended (1.2.0)
Owner

sferik commented May 23, 2012

Here's the full diff between 2.3.0 and 2.4.0: jnunemaker/twitter@v2.3.0...v2.4.0

I don't see anything in there that could be leaking memory but let me know if you see things differently.

You mentioned that you're not using the gem heavily. Specifically, what methods are you calling and how are you calling them?

modsaid commented Jun 3, 2012

I see nothing suspicious in the diff
My usage is: creating a list, obtaining the list_timeline, adding and removing members to the list
but most of them are from my background jobs server

I'm a bit clueless, but unable to try it again for the time being

Owner

sferik commented Jun 6, 2012

Can you let me know if the problem still existing in version 2.5.0? If that doesn't fix it, could you try 3.0.0.rc.1? I've made a number of performance enhancements.

@sferik sferik closed this Jun 9, 2012

modsaid commented Jun 10, 2012

Thank u @sferik, I am unable to try this at the time being.. tried 2.4.0 agian using webrick in a side process and things were smooth... trying it (and the newer versions) on our production is expected to be done soon. i'll keep you posted

Thank you for your help

sferik added a commit that referenced this issue Sep 18, 2012

Disable identity map by default
The identity map was causing more issues than expected:
* #260
* #285
* #302
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment