- **Type**: defect --> enhancement
---
** [tickets:#218] Problems in switchover with PLM enabled and TCP as the
transport**
**Status:** unassigned
**Milestone:** future
**Created:** Wed May 15, 2013 09:06 AM UTC by Mathi Naickan
**Last Updated:** Wed May 15, 2013 09:06 AM UTC
**Owner:** nobody
This is just a placeholder ticket to be a reference for the discussions in
userlist:
http://list.opensaf.org/pipermail/users/2012-December/003966.html
During switchover, PLM's ImplementerClear? is failing with TRY again and that
strangely because of IMMD down detected...
Could be some hidden bug somewhere...
Got to be seen
Changed 5 months ago by anders
This rings a bell for me.
I have seen this symptom when MDS over TIPC was used with TIPC incorrectly
configured.
If you get IMMD down, then it means that the IMMND no longer has MSD
connectivity with the
active IMMD. The only reason for that *should* be that the IMMD and the SC it
is executing on
has gone down. This *should* lead to a failover etc.
But if you have problems with MDS connectivity, or at least the realtime
handling
of such connectivity, then the failover mey never happen and there will be no
new
active IMMD and the IMMND keeps replying with TRY_AGAIN since that is all it
can do.
Changed 5 months ago by anders
In OpenSAF 4.2, IMMND subscribes in redundant mode towards IMMD.
This is fixed in 4.3 by ticket #2446.
This means that in 4.2, IMMND is exposed to the suposedly transient
unavailability of the IMMD,
whereas in 4.3, the IMMND never sees this because it is buffered by MDS.
For all normal imm applications, this should be no problem.
But if PLM, as an imm client here trying to invoke saImmOiImplementerClear, is
also
involed as a participant in the switchover mechanism, then possilby PLM is
blocking
the switchover, because it is blocked by the implementer-clear, which is
waiting on
the switchover -> deadlock.
If this is the problem, then a workaround could be for PLM to do a
saImmOiFinalize
on then a initialize a new OI handle, instead of doing the ImplementerClear?.
The finalize of the handle will force a hard implementer-clear as a side effect.
The explicit saImmOiImplementerClear does a "soft" implementer clear allowing
the
handle to survive, which is perhaps not necessary for PLM here.
Changed 5 months ago by anders
Unsolved ticket #2821 has some similarities to this one.
Changed 5 months ago by mathi
attachment Issue.zip added
logs
Changed 5 months ago by mathi
attachment imm.xml added
imm.xml
Changed 5 months ago by mathi
cc [email protected] added
Thanks for your looking into this Anders. I have just uploaded the logs shared
by the user previously.
I'm also asking him to enable traces of IMMD and IMMND and share them before we
could choose a direction w.r.t this topic.
---
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.
------------------------------------------------------------------------------
Don't Limit Your Business. Reach for the Cloud.
GigeNET's Cloud Solutions provide you with the tools and support that
you need to offload your IT needs and focus on growing your business.
Configured For All Businesses. Start Your Cloud Today.
https://www.gigenetcloud.com/
_______________________________________________
Opensaf-tickets mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets
