Piranha: can't ping realserver when pulse daemon started.
Hi Folks,
RedHat rhel5.6
I'm trying desperately to get ipvs load balancing up and running using the RedHat Piranha utility.
I'll post my config files later in the message.
In order to simplify things, I've only got one LVS director running (no backup director), and I've only got one realserver running.
I can ping the realserver static ip from the LVS director when the pulse daemon is NOT started. About one minute after I start the pulse daemon, I can no longer ping the realserver static ip.
I can ping the realserver static ip from any other box on the subnet.
The LVS app requires that the LVS director be able to ping (or CONNECT) the realserver as a sanity check to make sure the realserver(s) is/are still alive.
/var/log/messages entries:
Jul 28 10:49:21 emclb98 pulse[11281]: STARTING PULSE AS MASTER
Jul 28 10:50:51 emclb98 pulse[11281]: partner dead: activating lvs
Jul 28 10:50:51 emclb98 lvs[11293]: starting virtual service Test_Web_Server active: 80
Jul 28 10:50:51 emclb98 nanny[11298]: starting LVS client monitor for 90.21.26.128:80 -> 90.21.26.32:80
Jul 28 10:50:51 emclb98 lvs[11293]: create_monitor for Test_Web_Server/emcds032 running as pid 11298
Jul 28 10:50:51 emclb98 avahi-daemon[3072]: Registering new address record for 90.21.26.128 on eth0.
Jul 28 10:50:51 emclb98 avahi-daemon[3072]: Withdrawing address record for 90.21.26.128 on eth0.
Jul 28 10:50:51 emclb98 avahi-daemon[3072]: Registering new address record for 90.21.26.128 on eth0.
Jul 28 10:50:51 emclb98 nanny[11298]: [ active ] making 90.21.26.32:80 available
Jul 28 10:50:56 emclb98 pulse[11300]: gratuitous lvs arps finished
Jul 28 10:51:03 emclb98 nanny[11298]: CONNECT to 90.21.26.32:80 timed out
Jul 28 10:51:03 emclb98 nanny[11298]: [inactive] shutting down 90.21.26.32:80 due to connection failure
Jul 28 10:51:15 emclb98 nanny[11298]: CONNECT to 90.21.26.32:80 timed out
The LVS director is host emclb98 (90.21.26.98)
The realserver is host emcds032 (90.21.26.32)
The Virtual (floating) ip is 90.21.26.128
Here is my lvs.cf:
[root@emclb98 ~]# cat /etc/sysconfig/ha/lvs.cf
serial_no = 74
primary = 90.21.26.98
service = lvs
backup_active = 0
backup = 90.21.26.99
heartbeat = 1
heartbeat_port = 539
keepalive = 30
deadtime = 90
network = direct
debug_level = NONE
monitor_links = 1
syncdaemon = 1
virtual Test_Web_Server {
active = 1
address = 90.21.26.128 eth0:1
vip_nmask = 255.255.255.0
port = 80
send = "GET / HTTP/1.0\r\n\r\n"
expect = "HTTP"
use_regex = 0
load_monitor = none
scheduler = wrr
protocol = tcp
timeout = 6
reentry = 15
quiesce_server = 0
server emcds032 {
address = 90.21.26.32
active = 1
weight = 100
}
}
I'm using the arptables_jf solution described in the RedHat LVS installation guide:
ifconfig eth0:1 90.21.26.128 netmask 255.255.255.0 broadcast 90.21.26.255 up
arptables -A IN -d 90.21.26.128 -j DROP
arptables -A OUT -d 90.21.26.128 -j mangle --mangle-ip-s 90.21.26.32
service arptables_jf save
chkconfig arptables_jf on
Does anyone see anything glaring?
Thanks,
Cerephim
|