On Thu, 2010-01-28 at 15:55 -0700, hj lee wrote:
> Hi,
> 
> I am using corosync-1.1.2 on CentOS 5.3. In corosync.conf man page, I
> read "The rrp_problem_count_threshold * rrp_token_expired_timeout
> should be at least 50 milliseconds less then the token timeout". I
> think the "less than" should be changed to "more than". Or both less
> than or more than may be OK.
> 
> If the rrp_problem_count_threshold * rrp_token_expired_timeout is less
> than the token timeout, then the link become faulty before token
> timeout. This disables the link before trying GATHER mode. Is it the
> intended design? Shouldn't the totem try GATHER mode before disabling
> a link by faulty?
> 
> Is this (making a link faulty) a part of RRP protocol?
> 

It has been several years since I originally implemented RRP, and
unfortunately don't have an immediate answer to your question regarding
timing constraints.  I believe the constraint is there to ensure that
the last valid token is delivered before marking a link faulty.
Unfortunately I don't recall the exact reasoning for this constraint.
I'll have to do some research to gather an answer.  

Marking a link faulty is part of the specification, which can be read
here:

http://rcsc.de/pdf/icdcs02.pdf



> Thanks
> hj
> _______________________________________________
> Openais mailing list
> [email protected]
> https://lists.linux-foundation.org/mailman/listinfo/openais

_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to