I set up a local test environment and eventually saw
(a) OutOfMemorys (but only after a longer period of time, with -Xmx64M)
(b) Not being able to terminate with control-c (seemed to be depending
on whether Derby was instantiated or not, but can't confirm that)
(c) unexpected Postal terminations with many users (100.000)
But I am too new to this and would not say that i reproduced exactly
what you saw. I have not been able to isolate problems to the DB-case.
I would like to go on testing especially with two things:
+ Switch between DB/file and see how these two compare
+ Be able to peek into Derby and see if there is really data written to
the DB :-) (Maybe derby gives some more info on resource consumption)
A few questions turned up while testing:
How many users are you testing (local/external)?
How much heap should be given to James when load-testing?
Just out of curiosity: Why is DB not used for all repos?
-- bernd
Noel J. Bergman wrote:
We appear to have some major new problems with Derby support. It fails in
entirely non-graceful ways, as shown by the attached. Worse, we cannot kill
the process except with kill -9.
This was produced using the attached configuration file and:
$ ./smtp-sink 192.168.1.6:2525 1 &
$ ./postal -m 15 -p 20 -c 1 -r 100 192.168.1.101 userlist-postal - >
test.log &
$ ./rabid -r 1 192.168.1.101 userlist-rabid -
The test runs for a few minutes, and then the problem surfaces. So it
appears that I can spool, store and retrieve e-mail for a few minutes, and
them boom!
Does anyone spot anything wrong in my configuration? Does anyone else have
time to be working on this? :-)
--- Noel
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
