4 Replies Latest reply on Oct 3, 2017 8:19 AM by Steven Wyns

    Have you noticed that appservers are hammering the DB on 8.9.0 in a MAS setup

    Steven Wyns

      Hi Guys,

       

      While investigating an issue with BMC Support we gathered some AWR reports on our database and noticed that we have a lot of new connections to the database. Knowing bladelogic uses connection pools we were triggered.

       

      In our case we have 6 Client appservers, 6 Jobservers and 6 NSH proxy servers ( actually 6 servers each running a Client, Job and NSH instance).

       

      Now we notice that each of these servers is connecting to the db, executing the query below, and then disconnecting:

      select scp.name,
      scpv.config_property_value

      from service_config_properties scp,
      service_config_property_value scpv, application_server_deployments asd

      where scp.service_id = scpv.service_id

      and scp.config_property_id =
      scpv.config_property_id

      and scpv.appserver_deployment_id =
      asd.appserver_deployment_id

      and asd.host_name = :1 

      and asd.deployment_name = :2 

      and scp.name = :3

       

      It results in about 360 connects per minute on the oracle DB. Supports says it's the heartbeat. Altough from the query I find it hard to believe since it lists out the settings and doesn't query the alive times on the application_server table. Besides that it doesn't seem to be healthy to always open and close the connection?

       

      from the docs:

      To specify an interval between heartbeats for an Application Server, enter the following command:

      set appserver ServerMonitorInterval #

      where # is the frequency with which an Application Server updates its own time stamp (that is, its heartbeat). When an Application Server updates its heartbeat, it also checks for the heartbeat of any remote Application Servers.

       

      which is set to 20 on our env.

       

      The question is: has someone else noticed this as well? Or is this a misconfig on our side?