I am working on updating the user documentation (README and Programmer's 
reference). Will send out for review once they are ready. Will also improve the 
Doxygen documentation a bit in the new code.

Hints for testing this feature:

This feature is all about removing the limitation of max 2 system controllers. 
Use the IMM XML tools to generate a cluster with more than two controllers, and 
make sure that /etc/opensaf/node_type is "controller" on all those nodes. We 
are still using the 2N red model for OpenSAF directors, so there will still be 
one active and one standby. The rest are spares, ready to take over in case the 
active or standby fails. Use the command rdegetrole to check which node is 
active and standby (as usual). The spares will be reported as QUIESCED.

Hints for reviewing the code:

Most of the changes to the services have to do with supporting the QUIESCED 
role, i.e. running as a spare. The directors on a spare system controller will 
start up and register with AMF, but other than that they will not do anything 
and just wait for a role change. When a role change happens, they will continue 
the start-up sequence and initalize MDS so that they become visible on the 
network.

The role determination in RDE has been revised a bit, but essentially it works 
in the same way as before. The only real difference is that the assumption of 
max two system controllers has been removed. It can no longer assume that there 
is only one other peer node, and therefore it always has to wait for the full 
peer discovery time-out period even if a peer node has been discovered. 
Aborting the peer discovery early is not possible since the number of peers is 
not known.

Another difference in RDE is that it only selects the ACTIVE node. You can 
argue that this was already the case before, but since there used to be only 
two system controller nodes it was implicitly selecting the STANDBY when it 
selected the ACTIVE node. Now there can be many system controller nodes and RDE 
does not choose which one of the should be standby. AMF selects the STANDBY 
node, according to normal procedure for the 2N red model.

CLMNA has been modified to trigger the election of a new ACTIVE system 
controller when there is neither any ACTIVE nor STANDBY controller in the 
cluster. It will trigger the election by setting the RDE role to ACTIVE. RDE by 
default starts up in the QUIESCED role on all nodes.


---

** [tickets:#79] Dynamically changing the opensaf node type 
(controller/payload)**

**Status:** review
**Milestone:** 5.0.FC
**Labels:** #439 #1170 
**Created:** Mon May 13, 2013 04:33 AM UTC by Nagendra Kumar
**Last Updated:** Thu Mar 03, 2016 09:42 AM UTC
**Owner:** Anders Widell
**Attachments:**

- [Support for Spare System Controllers in 
OpenSAF.odt](https://sourceforge.net/p/opensaf/tickets/79/attachment/Support%20for%20Spare%20System%20Controllers%20in%20OpenSAF.odt)
 (70.9 kB; application/vnd.oasis.opendocument.text)
- 
[spare_sc.diff](https://sourceforge.net/p/opensaf/tickets/79/attachment/spare_sc.diff)
 (116.6 kB; text/x-patch)
- 
[spare_sc_uml_test.diff](https://sourceforge.net/p/opensaf/tickets/79/attachment/spare_sc_uml_test.diff)
 (2.6 kB; text/x-patch)


Migrated from http://devel.opensaf.org/ticket/2101

This ticket and subsequent ticket 
https://sourceforge.net/p/opensaf/tickets/1170/ is to support
the following two use scenarios
a) flexible deployment based on special hardware requirements for more than two 
system controllers
b) multiple nodes failing simultaneously in a cattle-type of deployment

The following is the scope of this ticket:

* It will be possible to configure nodes termed as spares (to the system 
controller group of nodes).
* If one of the system controller nodes dies, another "spare" system contoller 
node (if there is any available) will be get an assignment (active / standby). 
This assignment happens without requiring the spare system controller node to 
reboot.

We can add support for configuring more than two system controllers. One node 
will get an active assignment, another node will get a standby assignment, and 
the rest of the system controller nodes will have no assignment for their 
OpenSAF 2N SUs (they will be "spares").

For locking/unlocking the middleware 2N SU, refer the patch floated for 
http://devel.opensaf.org/ticket/2023

Ticket #1170 will be supported subsequent to this ticket.


---

Sent from sourceforge.net because [email protected] is 
subscribed to https://sourceforge.net/p/opensaf/tickets/

To unsubscribe from further messages, a project admin can change settings at 
https://sourceforge.net/p/opensaf/admin/tickets/options.  Or, if this is a 
mailing list, you can unsubscribe from the mailing list.
------------------------------------------------------------------------------
Site24x7 APM Insight: Get Deep Visibility into Application Performance
APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month
Monitor end-to-end web transactions and take corrective actions now
Troubleshoot faster and improve end-user experience. Signup Now!
http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140
_______________________________________________
Opensaf-tickets mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

Reply via email to