|
All,
I'm running a custom TCP service on port 80 which fields byterange
requests to clients. When running LVS-DR with WLC scheduling, an
ipvsadm -L -n displays seemingly incorrect ActiveConn counts. When
loading the server with 500 clients I will only see a dozen at best
ActiveConn entries. This leads me to believe that WLC may not be
working correctly, or may be scheduling clients incorrectly.
Config excerpts:
ldirectord.cf:
quiescent=yes
virtual=x.x.x.1:80
real=x.x.x.2:80 gate
real=x.x.x.3:80 gate
fallback=127.0.0.1:80 gate
scheduler=wlc
persistent=600
netmask=255.255.255.255
protocol=tcp
checktype=ping
failurecount=3
Testing using RR scheduling works as advertised and I can confirm that
the DR is indeed working from parsing the server logs being generated
for connection requests.
Perhaps it is a misunderstanding of the connection state related to
ActiveConn on my part, or something else. Any pointers are
appreciated.
Cheers
Kon
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users
|
