On 8/25/06, Stefano Bagnara <[EMAIL PROTECTED]> wrote:
Bernd Fondermann wrote:
> if, after a long, long run-time (a few days) under constant load the
> OOM pops up, I think it is safe to say there is a leak.
Time is not an important variable in leak hunting.
IMHO "several days" means nothing if I'm looking for a leak: I instead
want to know how many emails have been processed, how the tipical mails
looks like (big attachments, many recipients...), what are the
configurations and the versions of every single component involved (jvm,
OS, jdbc drivers, jars, custom mailets, database version).
Stefano
Sure. The point I was trying to make is that growing memory
consumption on application ramp-up is normal and that it is
furthermore perfectly possible to have OOMs without having a memory
leak.
In practice, it _is_ important how you distribute memory load over
time, because GC is not constantly but discretly running, which
affects the way memory can be freed. But I think this is not important
at all for this issue here.
Bernd
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
