On Aug 9, 2010, at 20:10 , Anoop Bhat wrote:

> Hi,
> 
> We use centos here and for HA, we get piranha which I know is more difficult 
> to support/get help for on this list but I'm hoping that someone can help me 
> figure out why (only recently) we're seeing real servers go inactive from the 
> LVS perspective.
> 
> The real servers are all tomcat6 and running services that we use on 8080 
> HTTP.
> 
> Our lvs.cf shows
> 
>         send = "GET / HTTP/1.0\r\n\r\n"
>         expect = ".*"
> 
> Which should result in some sort of response (400 bad request) and it usually 
> does.
> 
> I'm wondering if we're testing it incorrectly or if there is a better way to 
> test it?
> 
> These are all VMs in a ESX cluster and we're using LVS-DR instead of NAT or 
> TUN.
> 
> My understanding is that if it's HTTP, the send/expect above should work just 
> fine.


You may have encountered a bug that I've reported 
4 years ago. I'm not sure if it has been fixed as
we've only tested Piranha for a few days and  have 
never used it since. Check the access and/or error 
logs of your tomcats to see how the requests are 
corrupted.

https://bugzilla.redhat.com/show_bug.cgi?id=199243

regards,

rainer
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/

LinuxVirtualServer.org mailing list - [email protected]
Send requests to [email protected]
or go to http://lists.graemef.net/mailman/listinfo/lvs-users

Reply via email to