TCP RST client if service unreachable

[mcristobal]

I need to configure the CSS to send a TCP reset to the client if the service becomes unavailable or goes down so the client restablishes the TCP session to an active service within the VIP or rule.

Can you point me to the correct info/best practices??

Thanks

MANUEL

[seilsz]

This feature was implemented in WebNS version 7.30.0.05. The command (under the content rule) is 'flow-reset-reject'. The reference is located here:

http://www.cisco.com/en/US/products/hw/contnetw/ps792/products_configuration_guide_chapter09186a00801ee80e.html#wp1038399

~Zach

[Gilles Dufour]

just a small remark.

This command only works if you lose the arp entry to reach the server.

So if the service is down, but you can still ping the server, there will be no reset.

Regards,

Gilles.

[mcristobal]

Thanks Gilles,

here is my issue,...:

The front end or application must launch 25 TCP persistent sessions to the back end and use these sessions to pass traffic back and forth faster that establishing a new TCP session each time.

If the service at the back end goes down ( via keepalive script), the front end must restablish another 25 persistent TCP session but this time, to another alive service.

I was thinking that the CSS could send a TCP reset to the applciation to kill the 25 sessions when the back end service becomes unavailable and therefore forec the application to launch another 25 sessions to the next back end service.

Any ideas how???

THANKS

MANUEL

[seilsz]

Was that the intended behavior? So this feature is best suited for backend services (servers) are directed connected to the CSS?

~Zach

[mcristobal]

yes,, thsi is the intended behavior..

The servers are not directly plugged in to the CSS, just in the same segment.

[seilsz]

Sorry ... I was really responding to the post from Gilles.

~Zach