I have opsview connected to pagerduty and it seems to be working fine. I have a question about how to best configure opsview. Up until now we were only using email alerting, and since we have now moved to pagerduty, we want to cut the noise and work more towards a cleaner opsview setup, something we should've done it long ago, but we never did.
The way I look at this is that Opsview can generate an alert which gets sent to pager duty, which will generate a pagerduty alert if it falls between the parameters specified. What I'm trying to do is filter the number of opsview alerts that get sent to pagerduty.
We normally have service groups which have serveral checks. As an example, think of a postgres database, where we have a check for the listening port and another for, say, transactions taking a long time. We'd like to be alerted by the first one immediately but, on the transaction example, only after 2/3 hours it's been in critical state. I initially thought that maybe this could be done by having a few notification profiles tied to pagerduty which alert on the 1st opsview notification, 2nd notification, 3rd notification, etc.
But, as it stands now, all postgres checks are within the same service group, and this prevents from having them (each of the service checks) using different notification profiles.So, I was thinking that maybe what it needs is a few different service groups depending on the importance of the checks.
My question, is this a valid approach? Is this a 'best practice' within opsview? If not, how would you recommend configuring such a scenario?