Le 14/12/2016 à 06:24, Jacopo Cappellato a écrit :
> On Tue, Dec 13, 2016 at 9:43 PM, Jacques Le Roux <
>
[hidden email]> wrote:
>
>> Hi Taher,
>>
>> I don't see the need to discuss about this, I have expressed facts, that's
>> all
>>
>> Jacques
>
> but we need more facts to identify the core cause of the crash and fix it
> :-)
> I think that Craig asked the right question: if there was enough memory to
> start the instances and run them for 12 hours then the memory, under the
> same load, should be enough to run them forever. If on the other hand the
> memory usage slowly goes up then there is a memory leak or something else
> that needs to be fixed.
> Better move this conversation to the dev list where we can do a thorough
> analysis together.
>
> Jacopo
>
We are near the limit, it could simply be that an action needed more RAM, let's wait to see if it reproduce
Then we can make a post mortem analysis
Jacques
Locked
