changing installation passwords

[kendallchenoweth]

tungsten as a product isn't important to the question. What I want to do is use a database replication tool to sync two nagios mysql database environments (for a HA production pair) and tungsten will provide the ability to keep the two local server databases always in sync. This will obviously have implementation when I perform an upgrade on the HA server because the database will already have been updated by the upgrade process on the first server, but none of the other upgrade steps will have been performed yet. What is the recommended way to upgrade the nagios xi server excluding the mysql database content updates?

[sreinhardt]

OK, I have to start with, this is a highly irregular setup from anything else we have seen, and we really do not known how this might effect the base nagios core and XI systems. We have some other ideas on how this can be done, with similar but different methods that we can discuss if you would like.

As for your present question, we think it can be done, and likely will cause few if any errors on upgrade. Specifically thinking of the soon to be 2014 release, there will be new database additions, but very few if any deletions, this should allow any replication to continue even on an un-upgraded system to work fine. The catch would be that generally table and column additions would be performed in an if ( ! $table ) create table; type statement, that is not to say that we couldn't miss one or two. If that happens we can certainly walk through that case. Otherwise doing the regular upgrade on both systems, really should not be an issue as far as we know(again has not been tested in any way that we know of).

If you would like to discuss my\our thoughts on HA systems and how it would best be approached based on past experiences, I am happy to do so, I just didn't want to make a long post even longer.

[kendallchenoweth]

Can you describe or provide a link to your prefered and/or supported HA environment? Do you have any HA environments that you recommend with database replication?

Thanks!

[sreinhardt]

Unfortunately if we have a doc, it is rather outdated. However if you would like to use database replication along the path to HA, I would suggest something along the lines of:

Nagios Server A: Main system
Nagios Server B: HA\slave system

Replicate the nagiosql database and postgres database from A to B. - This will replicate all configuration changes, short of adding flat files for core. Applying config or using the reconfigure nagios script on replication, if there are changes would likely be a good idea.

Have System A do all main checking, this then forwards all results to system B as passive checks. This will cause both systems to have nearly identical results, just with slightly different time stamps.

Here is where the replication may become an issue rather than a good thing. You really want system B to have all checks be in a passive check state with a freshness check rate slightly longer than the active check rate on system A. This is allows system B to automatically begin checking, with the same checks as system A in the event A goes down, or does not report to B within the freshness time range.

The one catch, would be identifying how and when to turn on notifications for system B, as we presently do not have a way to tell core to enable\disable notifications on the fly as far as I know.

The issue we see with directly replicating all databases between systems, is that system B would be constantly checking, and also constantly sending alerts, in addition to attempting to add its own check results to a replicated DB that might cause conflicts with both DBs and rrd files. Along this same note is why I do not suggest replicating the ndoutils db across systems(aside from possibly acknowledgements and notes), as that contains all of the check results among other things.

[kendallchenoweth]

all set for now. Thanks!