Andy
I do none of the options presented on that page. My method is to simply
have two completely independent Xymon servers, and configure all clients
report to both servers. Operators know that if they cannot web to one
Xymon server, they should try the other. There smart enough to do this!
In theory I could do some kind of HA on the webserver, but the effort
isn't worth the pay-off IMHO.
The Xymon configurations are checked into a subversion repository (over
ssh, with key authentiation) so that they can be easily replicated between
the two Xymon servers. An operator that changes a file (eg hosts.cfg)
commits it to the repository and then does "ssh other-server svn update
`pwd`" (actually it's scripted and does some sanity checks also, but that's
the gist of it).
The two Xymon servers list IPs of themselves and the other Xymon server for
XYMSERVERS in their *client* configuration (xymonclient.cfg), so that each
Xymon server is fully monitored by the other one. But each Xymon server's
*server* configuration (xymonserver.cfg) lists only itself, to avoid
messages loops and duplicate messages.
J
On 12 February 2013 20:49, Andy Smith <user-982f5f6d4d28@xymon.invalid> wrote:
Hi,
I'm trying to configure a pair of xymon servers each in a different
datacentre, following the approach 2 given in
http://en.wikibooks.org/wiki/System_Monitoring_with_Xymon/Administration_Guide#HA-WAN_2_approach
Is this still a favoured approach and if so, how do you avoid a loop with
client reports being endlessly forwarded between Xymon servers? Can anyone
point me in the correct direction or show me what the configuration for
xymond and xymonproxy would look like?
Thanks
--
Andy