[erlang-questions] Re: Reliable way to handle Java node crashs?

[Mikael Pettersson]

zabrane Mikael writes:
 > Hi guys,
 > 
 > Please, could someone help me understand this memory heap alloc failure of
 > the Erlang VM  ;-)
 > 
 > Regards
 > Zabrane
 > 
 > 
 > > On Jun 29, 2:09 am, zabrane Mikael <zabra...@REDACTED> wrote:
 > >> > Trying to restart the same failing test, I got (maybe) a useful info:
 > >> >
 > >> > beam.smp(89316,0xb14d1000) malloc:
 > >> > *** mmap(size=730857472) failed (error code=12)
 > >> > *** error: can't allocate region
 > >> > *** set a breakpoint in malloc_error_break to debug
 > >> >
 > >> > Crash dump was written to: erl_crash.dump
 > >> > eheap_alloc: Cannot allocate 583848200 bytes of memory (of type "heap").
 > >> >
 > >> > Abort trap
 > >> >
 > >> > 2010/6/29 zabrane Mikael <zabra...@REDACTED>
 > >> >
 > >> > > Now, my Java node is restared after each crash !
 > >> > > But for one of my message data, I got this strange Erlang error:
 > >> >
 > >> > > beam.smp(89316,0xb14d1000) malloc:
 > >> > > *** mmap(size=730857472) failed (error code=12)
 > >> > > *** error: can't allocate region
 > >> > > *** set a breakpoint in malloc_error_break to debug
 > >> >
 > >> > > Could someone tell me what happens to the Erlang VM?

It's all there: the Erlang VM asked the OS for more memory,
but the OS refused.  Presumably you don't have enough RAM or
swap in the machine.

The 730M allocation is a tad big, but not unreasonable, so
it's not necessarily a bug in the Erlang VM.

If you need more help than this, post a self-contained test case.

ps. please don't top-post