jacob.rief@tis.at wrote:
>
> Hello Alan,
> Heartbeat now does a good job for some month. However last day I had a big
> problem. It seemed as if one of the Ethernet interfaces went down,
> probably it was hardware or a driver failure. This caused a service failure
> for a few hours. Anyway, since Heartbeat does not check if the
> Ethernet-interfaces
> goes down, it did not turn over to the standby loadbalancing router.
> Since I can assing an IP-address to the switches, it would be a nice idea
> to ping
> these switches at regular intervals and, if that fails to turn over to the
> standby host.
> This would be a very useful feature since it can happen that someone
> unplugs the wrong Ethernet cable.
> I think that 'pulse' distributed with RedHat 6.1 is doing something like
> that.
>
> BTW: I added a feature to update automatically the ldirector-config files
> onto
> the second heartbeated host through scp (secure copy). I think it would be
> a good
> idea to also update the /etc/ha.d/haresoure file this way. What do You
> think about?
Right now you can do that with "Mon", and a shutdown of the heartbeat
subsystem. In the future, it should probably be done by heartbeat
itself, and then trigger node transitions when the router comes and
goes, just a "real" node does.
Thanks!
-- Alan Robertson
alanr@suse.com
>
> Hello Alan,
> Heartbeat now does a good job for some month. However last day I had a big
> problem. It seemed as if one of the Ethernet interfaces went down,
> probably it was hardware or a driver failure. This caused a service failure
> for a few hours. Anyway, since Heartbeat does not check if the
> Ethernet-interfaces
> goes down, it did not turn over to the standby loadbalancing router.
> Since I can assing an IP-address to the switches, it would be a nice idea
> to ping
> these switches at regular intervals and, if that fails to turn over to the
> standby host.
> This would be a very useful feature since it can happen that someone
> unplugs the wrong Ethernet cable.
> I think that 'pulse' distributed with RedHat 6.1 is doing something like
> that.
>
> BTW: I added a feature to update automatically the ldirector-config files
> onto
> the second heartbeated host through scp (secure copy). I think it would be
> a good
> idea to also update the /etc/ha.d/haresoure file this way. What do You
> think about?
Right now you can do that with "Mon", and a shutdown of the heartbeat
subsystem. In the future, it should probably be done by heartbeat
itself, and then trigger node transitions when the router comes and
goes, just a "real" node does.
Thanks!
-- Alan Robertson
alanr@suse.com