Red Hat Bugzilla – Bug 144821
clumanager goes bonkers if bonding configuration is broken
Last modified: 2009-04-16 16:16:02 EDT
Description of problem:
Clumanager gains/loses quorum consistently on one node.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Configure a bonded interface.
2. Connect one of the bonded interface cables to the WRONG network so
that ethtool shows that there's a link, but the traffic is dead-ended
(testing was done with a crossover cable).
3. Configure clumanager to use the bonded interface's IP as the member
4. Configure clumanager in 2-node mode and use an IP tiebreaker which
is only reachable via the bonded IP address's route.
5. Set permanent soft quorum with "cludb -p cluquorumd%allow_soft 1".
6. Start clumanager on that node.
Clumanager quorum forms/dissolves continually.
This is a configuration problem, not a clumanager problem. When
pinging the IP tiebreaker in this configuration, the following output
[root@magenta network-scripts]# ping 192.168.1.254
PING 192.168.1.254 (192.168.79.254) 56(84) bytes of data.
64 bytes from 192.168.1.254: icmp_seq=1 ttl=255 time=0.282 ms
64 bytes from 192.168.1.254: icmp_seq=3 ttl=255 time=0.336 ms
64 bytes from 192.168.1.254: icmp_seq=5 ttl=255 time=0.302 ms
64 bytes from 192.168.1.254: icmp_seq=7 ttl=255 time=0.278 ms
64 bytes from 192.168.1.254: icmp_seq=9 ttl=255 time=0.326 ms
64 bytes from 192.168.1.254: icmp_seq=11 ttl=255 time=0.463 ms