I'm not really involved in the management/monitoring side of things at
the moment, but I expect that it would be enough for just each Geronimo
instance to know the PID of it's jvm. How you get that into it is
another matter :-)...
As long as each instance can give you its pid it does not matter whether
you are slicing at e.g. "Cluster", "Box", "Network" or any other
grouping of Geronimo instances via your management panel. It makes it
available to standard JMX monitors etc. via MBean/GBean/XBean
(whatever!) attributes etc...
Writing pid info into a common place would restrict you to either only
the instances running on one box, or a shared file system (not something
everyone wants in production) or aggregating this info from a number of
remote dirs. It might be useful for a development cycle, but I would be
happier with a management solution that was not bounded by physical
constraints in this way.
Just my penniesworth,
Jules
Bruce Snyder wrote:
On 11/20/05, John Sisson <[EMAIL PROTECTED]> wrote:
PID file thoughts (long term enhancement)..
I noticed that catalina.sh in Tomcat 5.5.12 has the ability to store the
PID in a file where the file name is supplied the CATALINA_PID
environment variable.
I am wondering if geronimo's startup script could somehow get hold of
the domain and server names for the Geronimo instance to be started.
Geronimo's startup script could store the PID for Geronimo (when
Geronimo is started) in a file whose name is constructed using the
domain and server names and stored in a central location that Geronimo
knows how to find.
This would enable one to look in a directory and see all the PIDs for
Geronimo instances currently running. This would allow us to create
diagnostic/management scripts that can use this information. It would
also allow Geronimo itself to read the file to find the PID if we or any
third party wanted to provide any integration of operating system
information in the console (would only be available for operating
systems where PID information is available). Just thinking out loud...
When we are run multiple instances of Geronimo, where will users
configure the domain and server names that are used as part of the GBean
names?
I hazard a guess that this would be configured from the Console. But
the clustering software will need to be made aware of this information
somehow so it would need to be involved with the startup/shutdown
cycles. I'm going to CC the WADI dev@ list on this to get that team
more involved in this discussion.
Bruce
--
perl -e 'print unpack("u30","D0G)[EMAIL
PROTECTED]&5R\"F)R=6-E+G-N>61E<D\!G;6%I;\"YC;VT*"
);'
The Castor Project
http://www.castor.org/
Apache Geronimo
http://geronimo.apache.org/
--
"Open Source is a self-assembling organism. You dangle a piece of
string into a super-saturated solution and a whole operating-system
crystallises out around it."
/**********************************
* Jules Gosnell
* Partner
* Core Developers Network (Europe)
*
* www.coredevelopers.net
*
* Open Source Training & Support.
**********************************/
