Hi,
In the last few days, I have executed some more serious runs of Alpha3
on my dedicated test server.
(2 year old server with sufficient resources in terms of RAM, CPU and
disks. OS = Linux)
James was running in a Java 5 Sun-JVM with unaltered settings (RAM,
Threading, GC), except for activated JMX.
Two different repository flavors have been tested. Due to bug JAMES-505
an all-file-repository configuration has not been tested, but I will
with Alpha4.
It ran with default (out-of-the-box) configuration (which is mostly
dbfile) and an all-db configuration.
repository config | runtime | mail sent | max heap[B]| max conc. threads
------------------+---------+-----------+------------+------------------
james-default | 8 hours | 26.864 | 16.261.120| 47
all-db | 8 hours | 26.871 | 16.392.192| 47
james-default | 3 days | 241.754 | 17.956.864| 70
From the detailed data it can be seen that the gradient of the max heap
value curve is converging to zero.
Not a single message got lost.
No errors in the log files.
From the maximum allocated heap it is clear that my standard test load
is far from being maxed out (James is taking only <18MB out of 64MB).
I'm currently increasing both the number of messages and the max size of
single messages to set a new bottom line for the next test runs.
If a test using file repositories yields similar results and all results
still hold for Alpha4, we could consider it stable enough to become a
release candidate, shouldn't we?
Bernd
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
