[erlang-questions] GC trace data

Jachym Holecek <>
Mon May 5 17:23:33 CEST 2008


I'm using 'erlang:trace(Pid, true, [garbage_collection])' to benchmark
process memory usage and I'm not sure about the exact meaning of data
delivered with gc_start/gc_end events. Here's list of keys and how I
understand them:

       heap_block_size - Total heap size.
             heap_size - Size of used part of the heap.
             mbuf_size - Size of messages queued in process mailbox.
   old_heap_block_size - ??
         old_heap_size - ??
           recent_size - ??
            stack_size - Current stack size.

Does it sound sensible? Can someone help me fill in the blanks?

What I really care about is: 1. stack size (got that), 2. mailbox
size (got that), 3. size of used part of heap (could be 'heap_size'
or 'heap_size + old_heap_size' or something completely different?).

I'm running R11B in case it makes any difference.

Thanks in advance,
         -- Jachym

More information about the erlang-questions mailing list