Heapy (Ruby Heap Dump Inspector)

A CLI for analyzing Ruby Heap dumps. Thanks to Sam Saffron for the idea and initial code.

Add this line to your application's Gemfile:

gem 'heapy'

And then execute:

$ bundle

Or install it yourself as:

$ gem install heapy

Run with two inputs to output the values of today.dump that are not present in yesterday.dump

$ heapy diff tmp/yesterday.dump tmp/today_morning.dump
Allocated STRING 9991 objects of size 399640/491264 (in bytes) at: scratch.rb:24

Run with three inputs to show the diff between the first two, but only if the objects are still retained in the third

$ heapy diff tmp/yesterday.dump tmp/today_morning.dump tmp/today_afternoon.dump
Retained STRING 9991 objects of size 399640/491264 (in bytes) at: scratch.rb:24
# ...

Pass in the name of an output file and the objects present in today.dump that aren't in yesterday.dump will be written to that file

$ heapy diff tmp/yesterday.dump tmp/today.dump --output_diff=output.json
Allocated STRING 9991 objects of size 399640/491264 (in bytes) at: scratch.rb:24
# ...
Writing heap dump diff to output.json

Due to how the garbage collector in MRI manages objects on the Ruby heap, heapy diff may produce incomplete or incorrect diffs under the following circumstances:

1. Heap compaction. When compacting the Ruby heap either manually via GC.compact or with auto-compaction enabled, heapy cannot produce accurate diffs, because objects may move into different heap slots and will appear as newly allocated even if they weren't. In general, any mechanism that moves existing objects from one heap slot to another will invalidate diff reports. Always turn off compaction before taking the <after> heap dump.
2. Temporary allocations. Diffs might omit objects from an <after> dump if Ruby allocated them into heap slots that were previously occupied by different objects in the <before> dump, and which were then deallocated between both dumps. To minimize the chance of this happening, triggering a major GC three or more times between heap dumps can help tenuring survivors and thus stabilizing the heap prior to taking a diff.

Step 1) Generate a heap dump. You could do this manually. Or you can use a tool like derailed_benchmarks

Step 2) Once you've got the heap dump, you can analyze it using this CLI:

$ heapy read tmp/2015-10-01T10:18:59-05:00-heap.dump

Generation: nil object count: 209191
Generation:  14 object count: 407
Generation:  15 object count: 638
Generation:  16 object count: 748
Generation:  17 object count: 1023
Generation:  18 object count: 805
# ...

NOTE: The reason you may be getting a "nil" generation is these objects were loaded into memory before your code began tracking the allocations. To ensure all allocations are tracked you can execute your ruby script this trick. First create a file trace.rb that only starts allocation tracing:

# trace.rb
require 'objspace'

ObjectSpace.trace_object_allocations_start

Now make sure this command is loaded before you run your script, you can use Ruby's -I to specify a load path and -r to specify a library to require, in this case our trace file

$ ruby -I ./ -r trace script_name.rb

If the last line of your file is invalid JSON, make sure that you are closing the file after writing the ruby heap dump to it.

You can drill down into a specific generation. In the previous example, the 17'th generation looks strangely large, you can drill into it:

$ heapy read tmp/2015-10-01T10:18:59-05:00-heap.dump 17
    Analyzing Heap (Generation: 17)
    -------------------------------

    allocated by memory (44061517) (in bytes)
    ==============================
      39908512  /app/vendor/ruby-2.2.3/lib/ruby/2.2.0/timeout.rb:79
       1284993  /app/vendor/ruby-2.2.3/lib/ruby/2.2.0/openssl/buffering.rb:182
        201068  /app/vendor/bundle/ruby/2.2.0/gems/json-1.8.3/lib/json/common.rb:223
        189272  /app/vendor/bundle/ruby/2.2.0/gems/newrelic_rpm-3.13.2.302/lib/new_relic/agent/stats_engine/stats_hash.rb:39
        172531  /app/vendor/ruby-2.2.3/lib/ruby/2.2.0/net/http/header.rb:172
         92200  /app/vendor/bundle/ruby/2.2.0/gems/activesupport-4.2.3/lib/active_support/core_ext/numeric/conversions.rb:131

You can limit the output by passing in a --lines value:

$ heapy read tmp/2015-10-01T10:18:59-05:00-heap.dump 17 --lines=6

Note: Default lines value is 50

If you want to read all generations you can use the "all" directive

$ heapy read tmp/2015-10-01T10:18:59-05:00-heap.dump all

You can also use T-Lo's online JS based Heap Analyzer for visualizations. Another tool is HARB

After checking out the repo, run $ bundle install to install dependencies. Then, run rake spec to run the tests.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

Bug reports and pull requests are welcome on GitHub at https://github.com/schneems/heapy. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.

The gem is available as open source under the terms of the MIT License.