I can confirm that the problem still exists with 2.6.14.2. I did not
get a reply the last time I reported the configurations, so I'm resending.
Our system uses keepalived to manage web, ssh, ftp, telnet, and rsync
connections. The web connection is by far the heaviest, with about
300 active connections at the quietest period, going up to well over
a thousand at peak times. We use direct routing, using private IP
addresses (xxx.local, 10.2.xxx). First, here is the result of
ipvsadm on a 2.4 system at the moment:
IP Virtual Server version 1.0.12 (size=4096)
Prot LocalAddress:Port Scheduler Flags
- RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP publ.mcc.ac.uk:ssh wlc
- pascal.local:ssh Route 8 1 0
- euclid.local:ssh Route 8 1 1
TCP publ.mcc.ac.uk:telnet wlc
- pascal.local:telnet Route 8 0 0
- euclid.local:telnet Route 8 0 0
TCP publ.mcc.ac.uk:ftp wlc persistent 50
- boole.local:ftp Route 8 2 51
TCP farm.mcc.ac.uk:www wlc
- zethos.local:www Route 16 79 199
- cauchy.local:www Route 8 41 112
- pliny.local:www Route 8 40 116
- thales.local:www Route 8 40 119
- kepler.local:www Route 8 41 109
- euler.local:www Route 8 42 96
- brahe.local:www Route 8 40 116
TCP jfarm.mc.man.ac.uk:www wlc
TCP farm.mcc.ac.uk:rsync wlc
- briseis.local:rsync Route 8 0 0
Then, here is ipvsadm-save:
-A -t publ.mcc.ac.uk:ssh -s wlc
-a -t publ.mcc.ac.uk:ssh -r pascal.local:ssh -g -w 8
-a -t publ.mcc.ac.uk:ssh -r euclid.local:ssh -g -w 8
-A -t publ.mcc.ac.uk:telnet -s wlc
-a -t publ.mcc.ac.uk:telnet -r pascal.local:telnet -g -w 8
-a -t publ.mcc.ac.uk:telnet -r euclid.local:telnet -g -w 8
-A -t publ.mcc.ac.uk:ftp -s wlc -p 50
-a -t publ.mcc.ac.uk:ftp -r boole.local:ftp -g -w 8
-A -t farm.mcc.ac.uk:www -s wlc
-a -t farm.mcc.ac.uk:www -r zethos.local:www -g -w 16
-a -t farm.mcc.ac.uk:www -r cauchy.local:www -g -w 8
-a -t farm.mcc.ac.uk:www -r pliny.local:www -g -w 8
-a -t farm.mcc.ac.uk:www -r thales.local:www -g -w 8
-a -t farm.mcc.ac.uk:www -r kepler.local:www -g -w 8
-a -t farm.mcc.ac.uk:www -r euler.local:www -g -w 8
-a -t farm.mcc.ac.uk:www -r brahe.local:www -g -w 8
-A -t jfarm.mc.man.ac.uk:www -s wlc
-A -t farm.mcc.ac.uk:rsync -s wlc
-a -t farm.mcc.ac.uk:rsync -r briseis.local:rsync -g -w 8
Finally, here is the keepalived.conf file:
! Configuration File for keepalived
global_defs {
notification_email {
[EMAIL PROTECTED]
}
notification_email_from [EMAIL PROTECTED]
smtp_server mailrouter.mcc.ac.uk
smtp_connect_timeout 30
lvs_id abel
}
vrrp_sync_group VG1 {
group {
VI_1
}
}
vrrp_instance VI_1 {
state MASTER
interface eth0
virtual_router_id 51
priority 100
advert_int 1
authentication {
auth_type PASS
auth_pass
}
virtual_ipaddress {
130.88.203.156
130.88.203.157
}
}
virtual_server 130.88.203.155 80 {
delay_loop 6
lb_algo wlc
lb_kind DR
protocol TCP
real_server 10.2.32.13 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.14 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
}
virtual_server 130.88.203.157 80 {
delay_loop 6
lb_algo wlc
lb_kind DR
protocol TCP
real_server 10.2.32.3 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.6 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.9 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.12 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.15 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.4 80 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.29 80 {
weight 16
TCP_CHECK {
connect_timeout 6
}
}
}
virtual_server 130.88.203.156 22 {
delay_loop 6
lb_algo wlc
lb_kind DR
protocol TCP
real_server 10.2.32.5 22 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
real_server 10.2.32.11 22 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}
}
virtual_server 130.88.203.156 23 {
delay_loop 6
lb_algo wlc
lb_kind DR
protocol TCP
real_server 10.2.32.5 23 {
weight 8
TCP_CHECK {
connect_timeout 6
}
}