Hi, Assuming that your criteria is for preferentiality w.r.t role assignments and not about preferentiality w.r.t starting/existence of those processes(C,D) itself, praveen's suggestion could be clarified/summarized as below:
- All the processes(particularly C & D) could be modelled as belonging to seperate applications modelled in the 2N redundancy. - Use/Define SU ranks to those applications such that during initial cluster formation, the SUs within C and D get ACTIVE assignments preferentially on card 2. - Upon card re-insertion (after extraction), implement a logic in C,D such that the applications C and D trigerrs an AMF si-swap admin operation for their respective SIs. i.e. applications will trigger si-swap upon detecting(determined through AMF PG tracking and or CLM tracking) that the peer(apps on card2) is up. Cheers, Mathi. ----- [email protected] wrote: > Hi Praveen, > Thanks for the information, I assume by ‘reboot’ you mean restart the > process? > > thanks > — > tony > > On Feb 20, 2014, at 6:30 AM, praveen malviya > <[email protected]> wrote: > > > > > On 20-Feb-14 1:57 AM, Tony Hart wrote: > >> Hi Praveen, > >> > >> Thanks for the explanation. Maybe I can re-phrase the question in > terms of processes rather than SAF entities. I want to have processes > A,B and E run on card1 and C,D on card2 with the same failover/revert > behavior I described below. How might I sent up the mapping of > processes (each process performs a distinct function) to SG’s, SU’s > etc to achieve that? > > With current OpenSAF implementation, one can solution can be > imagined. > > 1)All processes will be modeled as a separate 2N application. > > 2)saAmfSURank can be configured to set initial distribution of > > assignments amond the SUs in the same application. > > 3)Component representing the process (sayA) will perform PG > tracking > > and will be aware of node leaving and joining the cluster. > > 4) As soon as a active component gets PG tracking callback for > other > > component got the standby role because of node, hosting that > > component(actually SU), reboot. > > it will perform si-swap for its SI to swap the role. > > 5) CLM can be used to monitor node related events. > > > > > > Thanks > > PRaveen > > > >> thanks, > >> — > >> tony > >> > >> On Feb 19, 2014, at 1:58 AM, praveen malviya > <[email protected]> wrote: > >> > >>> In 2N model only one SU (say SU1) will be active for all SIs and > another > >>> SU(say SU2) will be standby for all SIs. > >>> So even if 5 SUs are present, one SU1 will be active and another > SU2 > >>> will standby on behalf of all the SIs in the system. > >>> Other SUs are referred as spare SUs. > >>> > >>> Let us suppose SU1 hosted on card1 is active for SI1, SI2,SI3 > .... > >>> and SU2 hosted on card2 is standby for SI1,SI2, SI3.... > >>> If card1 reboots, then standby SU2 will become active on behalf of > all > >>> the SI1,SI2,SI3 .... > >>> and others pare SU3 will become new standby. > >>> Now if card1 again comes up SU1 will not be given any role on > behalf of > >>> all the SIs because SU2 is active and SU3 is standby for all > SI1,Si2,SI3.... > >>> > >>> > >>> Thanks > >>> Praveen > >>> On 18-Feb-14 6:25 PM, Tony Hart wrote: > >>>> 2N (saAmfSgtRedundancyModel=1) > >>>> > >>>> On Feb 18, 2014, at 7:17 AM, praveen malviya > <[email protected]> wrote: > >>>> > >>>>> On 17-Feb-14 6:27 PM, Tony Hart wrote: > >>>>>> Suppose I have two (identical) server cards in a system and I > have 5 SUs (A,B,C,D,E) that each have an active and standby instance. > By default all the active instances run on card 1 and all standby on > card 2. > >>>>> Please tell which redundancy model is being used? > >>>>> > >>>>> Thanks, > >>>>> Praveen > >>>>>> What if two of these (C and D say) are heavy CPU users when in > the active state and I’d like to have their active instances > automatically assigned to the other card from A,B and E. So card1 > hosts active instances for A,B and E and card2 hosts active instances > for C and D (the standby instances are oppositely assigned). > >>>>>> > >>>>>> Next if card2 is removed I’d like all the active instances to > revert to card1 (A,B,C,D,E). > >>>>>> > >>>>>> Then if card2 is re-inserted I’d like the original assignments > to be re-instanted (card1=A,B,E and card2=C,D). > >>>>>> > >>>>>> Similarly for card1 insert/re-insert. > >>>>>> > >>>>>> Is this possible with openSAF? > >>>>>> > >>>>>> thanks > >>>>>> — > >>>>>> tony > >>>>>> > ------------------------------------------------------------------------------ > >>>>>> Android apps run on BlackBerry 10 > >>>>>> Introducing the new BlackBerry 10.2.1 Runtime for Android > apps. > >>>>>> Now with support for Jelly Bean, Bluetooth, Mapview and more. > >>>>>> Get your Android app in front of a whole new audience. Start > now. > >>>>>> > http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk > >>>>>> _______________________________________________ > >>>>>> Opensaf-users mailing list > >>>>>> [email protected] > >>>>>> https://lists.sourceforge.net/lists/listinfo/opensaf-users > > > > > ------------------------------------------------------------------------------ > Managing the Performance of Cloud-Based Applications > Take advantage of what the Cloud has to offer - Avoid Common > Pitfalls. > Read the Whitepaper. > http://pubads.g.doubleclick.net/gampad/clk?id=121054471&iu=/4140/ostg.clktrk > _______________________________________________ > Opensaf-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/opensaf-users ------------------------------------------------------------------------------ Managing the Performance of Cloud-Based Applications Take advantage of what the Cloud has to offer - Avoid Common Pitfalls. Read the Whitepaper. http://pubads.g.doubleclick.net/gampad/clk?id=121054471&iu=/4140/ostg.clktrk _______________________________________________ Opensaf-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/opensaf-users
