Yes, the vrrp transitions are seen due the fact that a task, READOP is not yielding. So the Cpu gets busy and misses to sent the vrrp hello's which are supposed to be sent in 1 min. And when the vrrp slave sees no Hello from the Master and the hold down timer is exceeded, 3 sec, it assumes itself as the master.
The forecast release with the fix is 4.0.7.0 (Release date is May 7th)
Did you get a Change Request (CR) number from Nortel?
The Nortel reps I've talked to seem not to know about this and I'd like to point them in the right direction.
I'm running 4.0.6 and haven't seen any issues, although my CPU usage is fairly low (<10%).
OK, thanks... I feel a lot better after reading the CR.
The problem seems to require leaving JDM (or perhaps other SNMP management tools) querying at a reasonably high rate. I only use the CLI and MRTG for graphs (at 5 minute intervals) so it seems unlikely I'll be pushing the CPUs hard enough to cause it to trip up...
Nortel has put out a product bulletin on this.
If the link below doesn't work its document id 2006007018
It looks like they've narrowed the problem down to when chassis information is polled, things like temperature and fan status. I only poll them via MRTG every 5 minutes so I'm not too worried, but it might be of concern to anyone who leaved JDM open.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.