Hans, The OpenSAF implementation is non-conformant with AMF B.02.01 and hence wouldn't align 1:1 with the state model descriptions in B.02 as the state models changed significantly in the B.02 spec. The ideal handling of this particular situation according to the specification is the following:
- AMF wouldn't try to re-instantiate the component again and the SU to which the component belongs will also enter the INSTANTIATION-FAILED state. - This will in-turn convert the readiness state of the SU to out-of-service and so no SIs can be assigned to it. - The operational state of the component and the SU becomes disabled as a consequence of the component entering the INSTANTIATION-FAILED state. - The ADMIN action SA_AMF_ADMIN_REPAIRED should be invoked to clear the error on the component so that the component is again usable by AMF. The last step above is the answer to your original question. However in B.01 it was slightly different where when a node entered a cluster all its components were disabled while in B.02 it is the opposite hence there will be a difference in behavior. Also the SA_AMF_ADMIN_REPAIRED is not available for OpenSAF implementation of AMF. That's why its sometimes hard to reconcile all the differences. Thanks Sayan > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Hans Feldt > Sent: Tuesday, December 18, 2007 8:48 AM > To: Mathivanan Np-G19859 > Cc: [email protected] > Subject: Re: [Users] Operator intervention > > Thanks but... > > Mathivanan Np-G19859 wrote: > > >> Does the unlock operation trigger AMF attempts to instantiate the > >> component again? > > > > No, component is not instantiated again, because amf cannot > determine > > What state the component is in or what is the problem with his > > instantiation. > > My question was/is, what does the __operator__ have to do for > AMF to try instantiate the component again? > > >> Another example is when an application has restarted to many times > >> within the probation period, what happens then and how is > that error > >> cleared? > > > > The next level of escalation is done & the escalated recovery is > > performed. > > This is configurable to any of SU RESTART, SU FAILOVER, NODE FAIL > > OVER/FAST... > > If all automatic AMF recovery actions has failed, how does > the __operator__ repair the system? > > 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
