[ejabberd] eheap memory allocation crashes

Andre Pang ozone at algorithm.com.au
Wed Jun 7 10:35:39 MSD 2006


On 26/05/2006, at 10:10 PM, Mickael Remond wrote:

> * Andre Pang <ozone at algorithm.com.au> [2006-05-25 10:43:31 +1000]:
>
>> Hi all,
>>
>> I'm getting crashes with ejabberd/Erlang with the following error
>> message:
>>
>>   Slogan: eheap_alloc: Cannot allocate 153052320 bytes of memory (of
>> type "heap").
>
> Did you install the Erlang supervisor patch ?
> https://support.process-one.net/doc/display/CONTRIBS/Supervisor+- 
> +Performance+improvement+for+dynamic+workers
>
> I have seen this error happen without this patch.

Hi all, just to follow up on this thread: I think we've managed to  
solve the crashing issues.  Stupidly enough, we just added more RAM  
to the machine and increased the swap space, and now it's been  
humming away perfectly.  The cause of it was that Erlang was using an  
absolute ton of memory for some processes' message queues, because  
our stress testing clients were spamming the server with a massive  
number of messages per second.  The peak memory usage goes through  
the roof for a few seconds, but quickly comes back down as the  
processes recover and start dealing with the incoming messages properly.


-- 
% Andre Pang : trust.in.love.to.save  <http://www.algorithm.com.au/>





More information about the ejabberd mailing list