|
Hi,
I had a working configuration, but i don't know what happend to it,
worked fine then where was only one data node, so i think something
about ARP??
Now, when i shut down one node also nothing works, the browser brings a
timeout error message if i access the virtual ip.
Heartbeat and the MySQL section are working fine. Both web sections are
making problems.
For the Load Balancer and MySQL CentOS4.4 is used, for the Web Servers
Cobalt Linux 7.0 (2.4.x Kernel)
The sysctl.conf on the MySQL Servers has following settings :
net.ipv4.conf.eth0.arp_ignore = 0
net.ipv4.conf.eth0.arp_announce = 0
Normally that should be 1 and 2 to work with ldirectord i think, but it
works.
On the Cobalt Linux these options available aren't available. But as is
said it worked before with one node.
I am at a loss.
If anyone would have an idea id be very happy.
Greetings
Andreas Pankratz
------------------------------
My files :
------------
haresources:
------------
hostnamelb1.domain.com ldirectord::ldirectord.cf LVSSyncDaemonSwap
IPaddr2::web1vip/16/eth0/broadcast
--------------
ldirectord.cf:
--------------
checktimeout=10
checkinterval=2
autoreload=no
logfile="local0"
quiescent=yes
# web1
virtual=web1vip:80
real=web1real1:80 gate 5
real=web1real2:80 gate 5
fallback=127.0.0.1:80 gate 1
service=http
request="ldirector"
receive="alive"
scheduler=wlc
protocol=tcp
checktype=negotiate
# web2
virtual=web2vip:80
real=web2real1:80 gate 5
real=web2real2:80 gate 5
fallback=127.0.0.1:80 gate 1
service=http
request="ldirector"
receive="alive"
scheduler=wlc
protocol=tcp
checktype=negotiate
# MySQL (Replication) for SELECT Queries
virtual=mysqlvip:3306
real=mysqlslave1:3306 gate 5
real=mysqlslave2:3306 gate 5
fallback=mysqlmaster:3306 gate 1
service=mysql
login="load"
passwd="loadpw"
request="SELECT 1;"
protocol=tcp
scheduler=wlc
checktype=negotiate
-------------------------
Ifcfg on the real servers
-------------------------
DEVICE=lo:0
ONBOOT=yes
BOOTPROTO=none
IPADDR=thevip
NETMASK=255.255.255.255
|
