Re: DCD (dead connection detection) on Linux

From: Mark Burgess <>
Date: Thu, 31 Oct 2013 12:35:41 +1100
Message-Id: <>

Hi Amir,

are you using clusterware to start the database listeners automatically or starting them manually whilst logged into the RDBMS unix account with the AutoConfig generated env file sourced?

Reason I ask is that if you use CRS to start the listener you need to have IFILE links back into the AutoConfig generated network config files as follows:

In your RDBMS $ORACLE_HOME/network/admin/sqlnet.ora add the following entry (do this on all db nodes):


Without adding in the IFILE links to the AutoConfig generated network files CRS will not pick up the AutoConfig settings.

By the way - you will probably need to bounce the db listeners to have this take affect as well along with bouncing the R12 mid-tier as well if this is the solution.



On 31 Oct 2013, at 11:50 am, Hameed, Amir <> wrote:

> Folks,
> Linux - RHEL 6/U4
> Oracle E-Business Suite R12
> I have noticed that DCD is not working quickly enough in my database even if I set "SQLNET.EXPIRE_TIME= 1" on the database server. If I crash my Linux host running the http server while I have connections established to the application, I can see that dead connections in the database corresponding to the application connections take somewhere from 12-15 minutes to get cleaned up. I can see that the TCP keep-alive parameters on Linux are set as shown below:
> cat /proc/sys/net/ipv4/tcp_keepalive_time
> 7200
> cat /proc/sys/net/ipv4/tcp_keepalive_intvl
> 75
> Does anyone know what are all the parameters that need to be set to allow DCD work within 1-2 minutes?
> Thanks,
> Amir
> --

Received on Thu Oct 31 2013 - 02:35:41 CET

Original text of this message