I hope you weren't publishing events with "retention time"? Would be interesting if you could share the particulars of the tests that you performed.
- Mathi. > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Hans Feldt > Sent: Wednesday, November 07, 2007 2:38 PM > To: [email protected] > Subject: [Users] Using tools like valgrind on OpenSAF daemons > > We have seen a problem in the event service, looks like a > heap memory leak. After a few days of testing, ncs_eds dumps > core that shows a huge heap. Nearly 1GB! > > Anyway, my idea to find problems like this without changing > OpenSAF internals all the time: > - All OpenSAF daemons should be started by a script (not the > case today, e.g. ncs_eds) > - The script should check for a service specific environment > variable before starting the daemon and start the daemon in > that context. > - This is the same thing as today with the XTERM variable in > the scripts, the difference is that the variable name is > unique and can be inherited from the environment. > > For example: > - The ncs_eds_start.sh script would probe the environment for > EDSXTERM. > If that variable exist, start the ncs_eds daemon using that variable. > > Maybe some healthcheck timeouts need to be increased for a > debug run like this, I don't know yet. > > The changes are very small and makes it possible to easily run e.g. > Valgrind or other tools with a minimum effort. > > What do you think? > > Regards, > Hans > _______________________________________________ > Users mailing list > [email protected] > http://list.opensaf.org/maillist/listinfo/users > _______________________________________________ Users mailing list [email protected] http://list.opensaf.org/maillist/listinfo/users
