Hi all,

I like the idea. The name is cool. Karaf could benefit from a standalone 
monitoring tool.

My rather unstructured ideas:

For the policies I would recommend a „blacklisting approach“, so monitor all 
the bundles that are deployed and only exclude those on a „blacklist“ (maybe 
because they are not that important, etc). Whitelisting needs discipline (I 
usually do not have).
A kind of absence-pattern monitoring/alerting would be great: „For jmx-property 
xy there should be a change per hour (between 9-17), if not —> alert“
An interface to the camel-metric component would be great to cater for your own 
metrics not exposed by logging or jmx.
>From a business perspective the SLAs should be cluster/cellar-aware so that a 
>bundle needs to be running on one instance only to be counted as available. So 
>this should be configurable somehow.

Cheers, Thomas.

Reply via email to