Re: Out-of-Memory Handling in Long-Running / Server Applications

On Sunday 13 April 2008 13:56, Peter Booth wrote:
> Here's my two cents:
>
> Catching an OutOfMemory error is a waste of time, because by the tim
> eyou get that error all bets ar eoff an dthere is no guarantee that
> the VM is functional.

I guess I have to agree with this. You can catch it and, on a fairly
hit-or-miss basis, perform some recovery and continue on, but it's far
from a reliable technique and as such, pretty much worthless.


> Here's the pragmatic solution:
>
> 1.) ....
> n.) ...
> 9) Have a cup of tea.

Can I have coffee, instead?


This recipe is suitable only for situations in which the memory
requirements are well defined and have some sort of natural maximum and
you're just trying to optimize the heap limits.

In my case, problems vary considerably in their memory needs and the key
deployment environment is within a servlet container, which may be
configured with varying heap limits.

So for me the only acceptable solutions are those that protect from
uncontrolled failure modes and the potential undermining of the servlet
container's stability when memory is exhausted.

The solution pointed to by Burt Beckwith and outlined at
<http://www.javaspecialists.eu/archive/Issue092.html> seems like the
best approach.


Randall Schulz

===================================
This list is hosted by DevelopMentor®  http://www.develop.com

View archives and manage your subscription(s) at http://discuss.develop.com