Re: EM 13, CC12 agent use shared server connection

Thanks Niall  

I have to check if the database is configured for shared servers for default, thanks for the tip.

I have walk through doc 1415999 and I see a lot of contention for having only two dispatchers and not enough shared server but I was looking for a workaround since any minimal change on this database takes months!

So I was thinking on forcing the agent on dedicated connection but couldn’t find the connection string anywhere.  

Now I have to figure out why the agents are connecting via shared server if this is not the normal behavior.  

FJA   From: Niall Litchfield <niall.litchfield_at_gmail.com> Date: Tuesday, November 1, 2016 at 5:32 AM To: <correo_at_fjandrade.com>
Cc: ORACLE-L <oracle-l_at_freelists.org> Subject: Re: EM 13, CC12 agent use shared server connection  

Cloud Control doesn't use shared servers as far as I'm aware, see below is relevant cluster wide summary from one of our busy (well we think so) RAC databases. There are no virtual circuit waits and the shared server stats are zero in each of the individual AWR reports. I imagine it is possible that your database is configured to default connections to shared server, in which case you should work through doc 1415999.1 to see what is going on. Usually, insufficient shared servers and/or long running queries.  

Obviously, you haven't shared your troubleshooting up to now but it at least appears to be based on "I can only touch certain things, so I must use them" rather than "what's the root cause, how do we address this".  

Top Timed Events
Instance '*' - cluster wide summary

'*' Waits, %Timeouts, Wait Time Total(s) : Cluster-wide total for the wait event
'*' 'Wait Time Avg (ms)' : Cluster-wide average computed as (Wait Time Total / Event Waits) in ms
'*' Summary 'Avg Wait Time (ms)' : Per-instance 'Wait Time Avg (ms)' used to compute the following statistics
'*' [Avg/Min/Max/Std Dev] : average/minimum/maximum/standard deviation of per-instance 'Wait Time Avg(ms)'
'*' Cnt : count of instances with wait times for the event

Maybe someone had this issue before and can bring me some light.  

I have two OEM, one EM 13 and other Cloud Control 12.

Both of them connects via shared server to the database no matter what, since this is a heavy transactional RAC the virtual circuit wait event rise high and is the main wait event in the database.

When I put both OEM to work ( I´m migrating from 12c to 13c) the Virtual Circuit wait sky rocket.

I tried to fix this adding “(SERVER=DEDICATED )” just before de service name on the connection string in the monitoring and updating the agents for both EM13 and CC12 but this isn’t working. The agent keeps connecting via shared server.  

I can´t touch this database, so I can´t raise the shared pool/buffer cache or do other tuning in the database, I have to change the connection string, but I don´t find where in the agent software is this.  

EM 13.2.0.0 on RH 6.x Linux64 -> all clients RH 6.x, BDD 11.2.0.3.X -> agent 13.2.0.0

EM 12.1.0.5 on RH 5.x Linux64 -> same clients -> agent 12.1.0.5  

Thanks for your time.  

FJA          

-- 

Niall Litchfield
Oracle DBA
http://www.orawin.info




--
http://www.freelists.org/webpage/oracle-l