Keepalive Issue

Unanswered Question
May 17th, 2007
User Badges:

Hello all,


Wondering if anyone can help me here. I've a Cisco CSS 11500 7.20 Build 104. I've 5 WEB services configured on a CSS whereby at the same time nearly every night we have a service transition change on 1 service only going from ALIVE state, to DYING and finally DOWN. It stays in DOWN state for a short period of time but always comes back ALIVE by itself. This issue never happens during the day.


I've done an Ethereal capture where I think the issue might be. The HTTP/1.1 200 ok (text/html) is missing from the sessions when the issue happens.


Many thanks for any insight you maybe able to give to this issue. Where is the problem the server of the CSS.


If I can provide any further informaiton please let me know.


Regards,

Michael.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
joquesada Thu, 05/17/2007 - 04:54
User Badges:
  • Bronze, 100 points or more


Hi Michael,


If you have the service on the CSS configured to do an HTTP keepalive, it would wait a 200 OK from the server when it does the probe. If the server fails to send the 200 OK, then the expected behavior on the CSS is to move the service to DYING and then to DOWN state.


Taking an sniffer trace - as you've done - is the best way to see this behavior. Have you been able to determine what is happening with the application on the server at the time this occurs?


Given that this is happening at the same time every night, I would suspect there is a process on the server that stops the WEB application for a while, failing to respond to the HTTP keepalive from the CSS. Thanks!


Regards,


Jose Quesada.


michaelnolan Thu, 05/17/2007 - 06:07
User Badges:

Hi Jose,


Well the sniffer trace was usefull in determining this keepalive missing.


The systems guys tell me there system is ok but sure what's new there. At least I've soemthing to go back to them now.


I just wanted to be sure it was not the CSS causing the problem and I think it's clear from the sniffer trace that the source IP 192.168.11.116 is not returning the HTTP keepalive frame.(see attached a snapshot).


I wondered whether someone else had a similar issue and what the cause was.


Regards,

Michael.



joquesada Thu, 05/17/2007 - 06:23
User Badges:
  • Bronze, 100 points or more


Hi Michael,


Definitely the trace is clear, the application is failing to send the 200 Ok to the CSS. Let me know what the system guys say about it. Thanks!


Regards,


Jose.


michaelnolan Thu, 07/19/2007 - 01:04
User Badges:

Hi Jose,


From your experience would the problem be with the application on the Web server or the communication between the Web server and the application server.


The ASP page is it purely on the WEB server and no backend communication? Or is it thst once the ASP page on the WEB server is polled by the CSS, there is a communication to the backend APP server from the WEB server.


Have you ever came across some root causes, as the application team and not finding anything.


Many thanks,

Regards,

Michael.

Actions

This Discussion