RE: {bb} failover: all tests purple

[Philip Clark <bblist@cannae.uklinux.net>]

On Fri, 2006-05-12 at 14:54, Aaron Clasby wrote:
> I mod'd the bb-hosts and added BBNET on the line of the secondary BB
> server, so its now working better at least. Most of the items went to
> normal status but they do change to purple and back every so often.
> 
> I do have FAILOVERBOTH=TRUE and we don't use fping.sh.
> 
> Also, its not sending out notices either.

The failover operations (NET and PAGER) work in subtly different ways.

For the PAGER part, the software is running continuously but in a
non-operational state. That's why the BBPAGER should be specified
in bb-hosts against the secondary server and "bbwarn:" is set to
FAILOVER in bbwarnsetup.cfg. The failover script then periodically
tests the operation of the primary BBPAGER, by attempting to send
a dummy alert message, and switches paging to an operational state
if that fails.

For the NET part, the software (bb-network.sh) does not run, under
normal conditions. The failover script tests for the presence of the
primary BBNET using "ping" and starts the bb-network.sh script if
that fails.

This produces a couple of apparent inconsistencies. Firstly, the
bb-hosts file requires the BBPAGER keyword added to the line for the
secondary server and the BBNET keyword omitted (otherwise the network
script will be started with BB and not in response to a failover).
Secondly, if BB is shut down on the primary (but the machine remains
visible  on the network) paging will failover, but the network tests
will not. It was for this reason that the FAILOVERBOTH setting was
added to the script. It forces the network tests to failover when
the paging does, even if the primary is still up.

So, if you just want BB to work for now, you could back out of the
changes made for the failover script and just run the secondary as
a normal server. Or, if you want to sort out the failover stuff
now, you should take the BBNET keyword out of bb-hosts again and
restart.

Next confirm that "bbd" and "failover" (but not bb-network) are
run at start-up. Then, FAILOVER-BBNET and FAILOVER-BBPAGER files
should be created in BBHOME/tmp to show that the failover has been
triggered. If those files exist before starting, either delete
them or pay attention to the timestamps to ensure that they are
re-written.

Cheers, Phil.



-- 
Mollison's Bureaucracy Hypothesis: If an idea can survive a
bureaucratic review and be implemented it wasn't worth doing.

--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=
To unsubscribe from this list, or to subscribe to the bb-digest list
send e-mail to mailto:majordomo@bb4.com with unsubscribe bb -and/or-
subscribe bb-digest in the BODY of the message.