[OO Tip] Keeping TCP connection alive when terminated by load balancers

[JasonCantrell]

Problem:

After an hour of inactivity the Secure Shell (SSH) connections are terminated by the load balancers. If a script is running on the target server this will also terminate. Is there a setting to keep the SSH session alive?

Solution:

Please contact OO Support for assistance with this question as there is a suggested workaround but not a certified solution. Please mention this KB article.

Link to the document here:

http://support.openview.hp.com/selfsolve/document/KM1456301

[jphaag]

Hello,

I encountered the same problem, but it concerns in my case tcp connection.

A Bladelogic job is launched by a ras? and when the RAS receive the answer of Bladelogic, He tries to return the response to Central, but the LB has terminated the session.

How to force Central to send a TCP keep alive probe to inform LB not to terminate the session considered as inactive by LB ?

What was the workaroud suggested for this case ?

Thanks for your answer.

Jean-Philippe HAAG

[molinabro1]

A suggested workaround would be to add parallel steps where the command is being executed in one lane.  The other lane is using a loop to send a bogus command to the server (like: echo test > /dev/null) to keep the connection alive.

For this to work, the closeSession input needs to be set to false in all iterations of SSH Command on the two lanes.

Steps are as follows:

1. SSH Shell logon
2. Parallel Split Step
3. SSS Command
4. SSH Shell Logoff