Hi Hans, It is not really a timer based. The determination by the payload that the system controller Did really go down has to use a timer to factor in communication glitches and recovery scenarios. So we will have a time factor but how long should be tuned for each platform based on their Network reliability and recovery times.
Regards Sugadeesh > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Hans Feldt > Sent: Thursday, November 22, 2007 1:10 PM > To: Kumar Nagendra-G20235 > Cc: [email protected] > Subject: Re: [Users] System behavior when > bothcontrollerstemporarily disappears? > > Thanks, now I understand. > > It would be nice to have a fool-proof solution for non-HPI > enabled systems. > > The payload node should reboot itself automatically but not > timer based, rather based on an event like both active and > standby controller missing. > > Thanks, > Hans > > Kumar Nagendra-G20235 wrote: > > If your system is not HPI enabled then user intervention is > required. > > You need to reboot the payloads manually. > > > > The case you are specifying would be solved on HPI enabled > system like > > this: > > When controller comes up then AVM of AvSv gets all the payload > > information from HPI using HISV and send a reboot command to reboot > > these payloads. So, all the payloads would be rebooted. > > > > Regards > > -Nagendra > > > > -----Original Message----- > > From: Hans Feldt [mailto:[EMAIL PROTECTED] > > Sent: Wednesday, November 21, 2007 5:41 PM > > To: Kumar Nagendra-G20235 > > Cc: [email protected] > > Subject: Re: [Users] System behavior when both > controllerstemporarily > > disappears? > > > > And this timer is canceled if an active controller comes > back before > > the timer expires? > > > > So if both controllers are back again within 3 minutes, the > payloads > > will not reboot, is that correct? > > > > But at the same time the payloads are not allowed to enter > the cluster > > again and we get the the AVD logs filled with: > > > > AN INVALID DATA VALUE at ./avd_ndfsm.c:346 val 132367 > > > > Please explain how this is suppose to work and if there is a catch > > here with the scenario described above. > > > > Regards, > > Hans > > > > > > Kumar Nagendra-G20235 wrote: > >> Hans, > >> The wait timer is in mds. MDS delays the DOWN events of > >> Directors(working on Vdest) by 3 minutes. The reference is > >> opensaf/services/mds/src/mds_main.c : uns32 > MDS_AWAIT_ACTIVE_TMR_VAL > >> = > > > >> 18000; AvND on payload doesn't get AVD down event till 3 > minutes. And > >> when it gets, it reboots. > >> > >> This time had been put large enough to make Switchover success. > >> Swithover may take some time in putting the components STDBY from > >> ACTIVE. > >> > >> You can fine tune it as per your requirements as far as payload is > >> concerned. > >> > >> Regards > >> -Nagendra > >> > >> -----Original Message----- > >> From: [EMAIL PROTECTED] > >> [mailto:[EMAIL PROTECTED] On Behalf Of Hans Feldt > >> Sent: Tuesday, November 20, 2007 8:18 PM > >> To: [email protected] > >> Subject: Re: [Users] System behavior when both > controllerstemporarily > >> disappears? > >> > >> I found avnd_di.c:avnd_evt_mds_avd_dn that does a reboot directly > >> from > > > >> what I understand. But the payload node seems to wait 3 minutes > >> before > > > >> reboot. Where is that in the code? > >> > >> Thanks, > >> Hans > >> > >> Hans Feldt wrote: > >>> We discussed this a bit in thread: > >>> http://list.opensaf.org/archives/users/2007-August/000108.html > >>> > >>> I would like the payloads to discover that there is no active > >>> controller and reboot themselves. Where in the (avnd) code can I > >> change this? > >>> This is a system without HPI enabled. > >>> > >>> Thanks, > >>> 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 > _______________________________________________ Users mailing list [email protected] http://list.opensaf.org/maillist/listinfo/users
