From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.0; Windows 98; DigExt)
I am using Red Hat 6.2 upgraded to HA 1.0. When the inside network NIC is
physically unplugged on the primary LVS router, the virtual services are
not transfered to the backup LVS router. All that happens is that the
Real Servers are lost by the LVS Primary Router and web services go away.
However, all is fine when the public (outside NIC) is unplugged.
Steps to Reproduce:
1.Unplug the Inside NIC from the hub on the Primary LVS router.
Actual Results: Lost the Real Servers in the Primary LVS routing table as
viewed on piranha. The web site is DOWN.
Expected Results: The pulse service releases the virtual server and nanny
checking stops. All relevant confiuration information is started on the
backup LVS router and browsing continues. This is what happens if the
outside NIC is unplugged on the Primary LVS Router.
Sorry that I have resorted to calling this a bug, I have been at this for
quite some time trying to find my configuration fault. But since when the
outside NIC is unplugged, all the services transfer to the backup LVS
properly, then I think that there is something else going on here. I
believe that the LVS backup should be checking both the internal and
external NIC's on the primary LVS router.
>I am using Red Hat 6.2 upgraded to HA 1.0.
What do you mean by this? If you used the UPDATE option of the HA product on an
existing Linux system, you do not end up with a proper HA server. The
documentation mentions this. The UPDATE option just updates your 6.2 system to
>All that happens is that the
>Real Servers are lost by the LVS Primary Router and web services go away.
> However, all is fine when the public (outside NIC) is unplugged.
When a testing a defined virtual service resulting in no real server responding
(i.e. all of the servers in that service are gone), it should failover.
Can you make sure you are running the latest software? Both IPVS and piranha
have changed quite a lot since the HA 1.0 release. This includes changes to I/O
testing and fixing a memory leak. Visit http://ha.redhat.com/ and grab the
latest RPMs from the experimental area, and see if you still have this problem.
Humm,... I'll do a bit of work on this and see whats up
I think I've got this fixed with 0.7.0-3
please note that te failover web pages were removed but the internal
functionality for failover in piranha still exists. (This was done for the AS
release of RHAT Linux where I was asked to remove the failover pages for