Working with someone who claims that four Solaris 10 servers didn't
update their ARP caches, after a power outage effected some of their
load balancers. So, the hosts continued to use the MAC addresses of the
wrong load balancers. To clear the problem up, they would just delete
the old MAC address, via 'arp -d', and the correct one would get added.
Of course, we don't have any snoop type data to show that the hosts were
actually given the correct MACs, via GARP, etc. We've skimmed through
100's of S10 ARP bugs, and did find a close match in CR 6690022 (IPMP
doesn't recover on router failover as arp entry is not updated by GARP),
which is the only bug fixed in latest S10 ARP patch.
We've recommended they apply this patch, but I am wondering if there are
other related bugs on this situation. Has anybody else dealt with these
types of ARP issues? What was the resolution?
Thanks,
Jim
_______________________________________________
networking-discuss mailing list
[email protected]
