Mathivanan Np-G19859 wrote: > Hans, > > - To add on to matt's comments, > > Looks like in this case saEvtEventRetentionTimeClear() was not called > to clear the retained events.
You're right. The test app is not calling this function. Didn't not read the specs enough I guess... > If so, for the 'big' timer value that has been used in this case, these > events would have gotten 'retained' on the eds server (ncs_eds process) > until the timer expired for these events. > Hence, the big memory usage that you get to see which are nothing but > the 'retained events'. Yes but what is important is that ncs_eds crasch and dumps core due to this! My point is that an OpenSAF server should never crasch due to misuse from a client. Could perhaps be called a Denial Of Service attack? Regards, Hans > > Rgds, > Mathi. > > > > > > >>-----Original Message----- >>From: [EMAIL PROTECTED] >>[mailto:[EMAIL PROTECTED] On Behalf Of Schoen >>Matthew-MSCHOEN1 >>Sent: Thursday, November 08, 2007 8:05 PM >>To: Hans Feldt; Murthy E-G19462 >>Cc: [email protected] >>Subject: Re: [Users] EVT problem >> >>After the subscriber receives the event is it invoking the >>retention time clear SaEvt API? >>-Matt >> >>-----Original Message----- >>From: [EMAIL PROTECTED] >>[mailto:[EMAIL PROTECTED] On Behalf Of Hans Feldt >>Sent: Thursday, November 08, 2007 7:28 AM >>To: Murthy E-G19462 >>Cc: [email protected] >>Subject: [Users] EVT problem >> >>Murthy E-G19462 wrote: >> >>>>-----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! >>> >>> >>>Please provide the leak details for our investigation. >> >>The retention time for a created event was set to >>SA_TIME_END. Events were produced at a rate of 4 per second. >>Either the consumer stopped consuming or the events were sent >>out in the blue. >>Eventually the 'ncs_eds' process got out of heap (or something else). >>The core dump cannot be used by GDB. >> >>That's all we know. >> >>We will change from SA_TIME_END to something else. >> >>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 >> > > _______________________________________________ Users mailing list [email protected] http://list.opensaf.org/maillist/listinfo/users
