2 Replies Latest reply on May 9, 2019 9:39 AM by aiglee.castillo

    Custom alerts on repository

    aiglee.castillo

      Hi,

       

      I am trying to figure out an availability alert that will work for us and nothing seems to work as expected. We are monitoring 15 DBs from the same DPA instance. I want to be notified if we have issues accessing any of the DBs.

       

      I tried the "Database Instance Availability" alert, and even though it did notify me, I want something that will notify me if the DB has been unavailable for X time, not at the one second the alert checked. The way it notifies currently it gives me too many false positives.

       

      I also tried an alert I found in the forums that can be run against the repository but it doesn't work as expected:

      select NAME+' Monitor is stopped', 1 from [ignite].[COND] WHERE [STATUS] = 'STOPPED' and [COMMAND] = 'START'

      The query is fine but if I select all DB instances and only one has the issue, you receive 15 emails. If I don't select all instances, it doesn't run. Why is that? All my alerts against repository need to have all DBs selected.

        • Re: Custom alerts on repository
          jaminsql

          The Database instance availability alert that is built in has some added parameters to it that are not very obvious that you can try changing. They are found in the options menu and described in this article.Success Center Most the time the issue is not that it give false positives but, that it doesn't go off as much as a user expects because or those settings.

          The monitor has to be on but, connection to the server is not working. Then we also have a tolerance in the code where we sample every 90 seconds by default and if we have 3 attempts that fail 90 seconds apart we alert.

          I would try tuning the items listed for attempts and seconds apart and see if you can find a good balance for your servers.

          As for the THWACK alert for a custom alert to run it has to be assigned to a monitor and the monitor must be started. For those reasons the above code is problematic at best for what it is trying to do. You could add an and clause to only alert on one instance such as "select NAME+' Monitor is stopped', 1 from [ignite].[COND] WHERE [STATUS] = 'STOPPED' and [COMMAND] = 'START' and Name='someServer'; The issue is you have to have someServer be any other server but, the one that you select in that alert and you also have to then have a version of the alert on each instance and have them pointed to some other server.

          Like server A alerts on Server B or something. Also Server A can't be down when server B is down of you also will not get an alert. This would really all be fixed if you could alert on the repository without having to select any monitored instance. There is a feature request that fits that same use case here.

          Run custom alerts against the repository without having to choose an arbitrary instance

          Perhaps upvote that item.

          I would say at this time the best you can do is use the built-in alert and if you see false positives tune the advanced options.

          1 of 1 people found this helpful