On 12-Nov-14 7:31 PM, Hobby Tan wrote: > Set saAmfSgtDefAutoRepair=0 will also disable low level recovery like > failover only > There are 2 parameters saAmfNodeSuFailoverMax & saAmfNodeSuFailOverProb can > change it, but it is node level instead of SG level. > > So I want to know if there is anyway to clean the Failover counter or > component restart counter to avoid it escalate to higher level? > If fact, what I want is SU can failover but don't trigger node reboot. >
If saAmfSgtDefAutoRepair=1, then su will be repiared by AMF. But this can be lead to node failover escalation if some components in the SU continuously faults. So user needs to decide whether repaired should be done by AMF or admin. Regarding the counters, they can be reset in one way: keep all the timers as low as possible. I think, not sure, we have a lowest limit 300ms. What is the impact of keeping them low? After expiry of any probation timer associated with any escalation level, AMF resets the counters associated with that level and freshly counts the faults. In this way keeping timers low will increase the probability of them being reset by AMF which eventually will reduce the probability of next level escalation. Thanks Praveen > Thanks, > > Hobby > > > -----Original Message----- > From: praveen malviya [mailto:[email protected]] > Sent: Wednesday, November 05, 2014 5:20 PM > To: Hobby Tan; [email protected] > Subject: Re: [users] AMF Escalation: Don't want the application recovery will > escalate to node reboot > > > > On 05-Nov-14 2:28 PM, Hobby Tan wrote: >> Dear all, >> currently we have several applications running in SC blade, I want to >> control one application only failover between SC-1 and SC-2 when it is >> failed to start, and then after trying several times(example: 10 times >> failover), it will stop trying and stay in failed state. But current >> behaviors will always trigger Node reboot finally. And I want to find >> which amf attributes are used to control this escalation. But this >> changing shall not impact other applications escalation to Node >> reboot. My application is 2N mode. and the related configuration is >> as below, I found Node level configuration would be the root caue of >> Node reboot, so my question would be how my application will not >> trigger it. Thanks for your support Object >> > Please try to configure recovery policy as sufailover. It can be done in two > ways: > 1) saAmfCtDefRecoveryOnError=2, saAmfCtDefDisableRestart-1, > saAmfSutDefSUFailover=1, saAmfSgtDefAutoRepair=0. or > 2) saAmfCtDefRecoveryOnError=3, saAmfSutDefSUFailover=1, > saAmfSgtDefAutoRepair=0. > > When suFailover will be escalated for the faulty SU, AMF will clean up all > the components of this SU and SU will remain in disabled state and there will > not be further escalation to node reboot. Since sg auto-repair is not > allowed, admin repair op will be required to enable the SU. > > Thanks, > Praveen >> Configuration >> >> Value >> >> safAmfNode >> >> saAmfNodeSuFailoverMax >> >> 2 >> >> saAmfNodeSuFailOverProb >> >> 1200sec >> >> saAmfNodeFailfastOnTerminationFailure >> >> 0 >> >> saAmfNodeFailfastOnInstantiationFailure >> >> 0 >> >> saAmfNodeAutoRepair >> >> 1 >> >> SaAmfSGType >> >> saAmfSgtDefCompRestartProb >> >> 60 sec >> >> saAmfSgtDefCompRestartMax >> >> 1 >> >> saAmfSgtDefSuRestartProb >> >> 60sec >> >> saAmfSgtDefSuRestartMax >> >> 1 >> >> SaAmfSUType >> >> saAmfSutDefSUFailover >> >> 1 >> >> SaAmfCompType >> >> saAmfCtDefClcCliTimeout >> >> 10sec >> >> saAmfCtDefCallbackTimeout >> >> 60sec >> >> saAmfCtDefQuiescingCompleteTimeout >> >> 60sec >> >> saAmfCtDefRecoveryOnError >> >> 2 >> >> >> >> >> >> >> >> >> ---------------------------------------------------------------------- >> -------- >> >> >> >> _______________________________________________ >> Opensaf-users mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/opensaf-users >> ------------------------------------------------------------------------------ Comprehensive Server Monitoring with Site24x7. Monitor 10 servers for $9/Month. Get alerted through email, SMS, voice calls or mobile push notifications. Take corrective actions from your mobile device. http://pubads.g.doubleclick.net/gampad/clk?id=154624111&iu=/4140/ostg.clktrk _______________________________________________ Opensaf-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/opensaf-users
