|
We are running ldirectord wlc mode persistent 1800
with 4 similar apache real servers. We've been
receiving complains from customers saying sometimes
that their sessions were timedout. I tried simulating
the problem. During my testing, my http requests were
severed by on of the real servers and i was seeing the
requests in the access_log. After about 15 minutes
browsing a shopping card of one of our customers, all
of sudden my browser timedout. During the timeout, my
HTTP requests did not reach the real webserver. I
checked the load balancer and the active connections
were around 20 connections for all 4 servers. On
average, active connections are 300. I run netstat -i
and the output was:-
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK
TX-ERR TX-DRP TX-OVR Flg
eth0 1500 0488803974 0 0 0858406697
0 0 0 BMRU
eth1 1500 0890907123 0 0 0514353823
0 0 0 BMRU
RX-OVR is 858406697 for eth0 which is VIP and
514353823 for the interface to the real servers
network.
The ipvs sysctl param are as follows:-
drop_packet 0
amemthresh 1024
sync_threshold 3
cat drop_entry 0
drop_packet 0
secure_tcp 0
amemthresh 1024
am_droprate 10
Any advise on why ldirector behaving like this.
Regards,
__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus
|
