Can I capture CSS keepalive response times?

Unanswered Question
Aug 7th, 2007
User Badges:


Does anyone know if it's possible to see the elapsed time taken for the latest keep-alive response for a given service?

Presumably the CSS knows it (since if it's less than 2 seconds before the next one is due to be sent the service is flagged as dying) but I've had a look in the Content Config Guide ( and can't see anything relevant.

I also had a look in the Admin Guide ( section about CSS MIBs but couldn't see anything either.

The reason I'm thinking about this is primarily that it would be useful for setting up an appropriate keep-alive frequency. However, in production the response time could also be monitored as an early warning sign of performance degradation (and alerted by your usual network management system).

Any comments/thoughts gratefully received!


  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
sadbulali Mon, 08/13/2007 - 10:30
User Badges:
  • Bronze, 100 points or more

Use the dns-boomerang client domain command to create a domain record in the Content Routing Agent DNS server for each of the domains you associated the agent with when you configured domains on the Content Router. If the matching domain record keepalive messaging succeeds, the CSS uses this record for DNS resolutions. There is no Content Routing Agent configuration mode

veriton Wed, 08/15/2007 - 02:05
User Badges:

Thanks for the reply Sadbulali.

However I'm a bit confused as to how that helps me tell how long the keep-alive response is taking to return from the server.

What I'm after is something that says that the keep-alive for a specific service took, say, 1420 msec (or whatever).

Any chance you could clarify?

veriton Mon, 09/24/2007 - 01:28
User Badges:

Anyone else reading this conversation might be interested to know that F5 Big-IP load balancers have the facility to set the maximum response time that a keep-alive ("performance monitor" in Big-IP terminology) is allowed to take and this is independent of interval between the keep-alives. E.g. you can say the keep-alive must respond within 5 seconds but only run it every 30 seconds.

In comparison on the CSS (v7.20 on) the timeout is the frequency minus 2 seconds, e.g. for a default 5 second frequency the timeout is 3 seconds.

Just an idea of what *might* be possible in future CSS versions...

I still think it would be useful to have the last keep-alive response time exposed via SNMP - I doubt it would add any/much additional load to the CSS and the more information you have the better IMHO.

Hopefully someone in Cisco CSS development might be reading this...


This Discussion