|
Hi all,
I am having a HA setup with two load balancers running ldirectord talking to
two dns servers using direct routing:
10.0.1/24 192.168.1/24
|-[DNS1]-|
->---|-[LB1 ]-|
->---|-[LB2 ]-|
|-[DNS2]-|
I am testing by sending DNS queries using queryperf to the VIP from outside
the subnet.
By quickly failing a DNS server (stopping bind9) while it was the active
server, I was able to have a UDP connection in the connection table that
never expires.
The timer keeps counting down to 0, then starts again from 60.
lb2:/etc/ha.d# ipvsadm -Lnc
IPVS connection entries
pro expire state source virtual destination
UDP 00:46 UDP 10.1.0.2:32799 10.10.10.1:53 192.168.1.22:53
lb2:/etc/ha.d# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
UDP 10.10.10.1:53 wrr
-> 192.168.1.21:53 Route 5 0 0
-> 10.0.1.22:53 Route 0 0 0
-> 10.0.1.21:53 Route 1 0 0
-> 192.168.1.22:53 Route 0 0 1
lb2:/etc/ha.d# sysctl -A|grep vs
net/ipv4/vs/nat_icmp_send = 0
net/ipv4/vs/sync_threshold = 3 50
net/ipv4/vs/expire_quiescent_template = 1
net/ipv4/vs/expire_nodest_conn = 1
net/ipv4/vs/cache_bypass = 0
net/ipv4/vs/secure_tcp = 0
net/ipv4/vs/drop_packet = 0
net/ipv4/vs/drop_entry = 0
net/ipv4/vs/am_droprate = 10
net/ipv4/vs/amemthresh = 1024
lb2:/etc/ha.d# ipvsadm -v
ipvsadm v1.24 2003/06/07 (compiled with getopt_long and IPVS v1.2.1)
lb2:/etc/ha.d# uname -a
Linux lb2 2.6.11.10.05052702 #1 SMP Fri May 27 18:17:56 EST 2005 i686
GNU/Linux
Queryperf uses the same source port for the queries, so when this happens
the queries are still sent to 192.168.1.22 even with the realserver marked
as not available.
Has anyone seen anything like this?
Regards
Andy
|
