LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Problems with ipvsadm connection synchronization

To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: Problems with ipvsadm connection synchronization
From: Sebastiaan Veldhuisen <seppo@xxxxxxxxxx>
Date: Fri, 03 Jun 2005 09:47:00 +0200
Hi,

we just implemented a 2nd director for our HA LVS environment and we 
want to do connection synchronization between the master and the backup 
director through ipvsadm

On both servers we added the multicast route:
Destination     Gateway         Genmask         Flags Metric  Ref    Use    
Iface
BASE-ADDRESS.MC *               240.0.0.0       U     0       0      0      eth0

On the master director we we started the synchronzation daemon:
ipvsadm --startdaemon=master --mcast-interface=eth0

On the backup director we started the synchronization daemon:
ipvsadm --startdaemon=backup --mcast-interface=eth0

Both directors are running the following versions of ipvsadm:
ipvsadm v1.24 2003/06/07 (compiled with getopt_long and IPVS v1.2.0)

We see multicasts of ipvsadm sync daemon:

master:~ # tcpdump multicast -i eth0 -n -p
listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes
18:37:12.084243 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 628
18:37:13.083984 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 28
18:37:14.083902 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 52

slave:~ # tcpdump multicast -i eth0 -n -p
listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes
18:34:37.105027 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 28
18:34:38.104901 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 100
18:34:39.104788 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 268
18:34:40.104692 IP x.x.x.x.45605 > 224.0.0.81.8848: UDP, length: 100

We also see the transfer of the connections on the backup:
slave:~ # ipvsadm -lcn | wc -l
1479

However the routing table remains emtpy on the slave director:
lummi:~ # ipvsadm -ln
IP Virtual Server version 1.2.0 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  x.x.x19:25 wlc
  -> x.x.x9:25             Route   1      0          0
  -> x.x.x8:25             Route   1      0          0
TCP  x.x.x18:80 wlc
  -> x.x.x5:80             Route   1      0          0
  -> x.x.x4:80             Route   1      0          0
TCP  x.x.x20:636 wlc
  -> x.x.x14:636           Route   1      0          0
  -> x.x.x15:636           Route   1      0          0
TCP  x.x.x21:636 wlc
  -> x.x.x7:636            Route   1      0          0
  -> x.x.x13:636           Route   1      0          0
TCP  x.x.x20:389 wlc
  -> x.x.x14:389           Route   1      0          0
  -> x.x.x15:389           Route   1      0          0
TCP  x.x.x21:389 wlc
  -> x.x.x7:389            Route   1      0          0
  -> x.x.x13:389           Route   1      0          0
TCP  x.x.x19:143 wlc
  -> x.x.x9:143            Route   1      0          0
  -> x.x.x8:143            Route   1      0          0
TCP  x.x.x18:443 wlc persistent 3600
  -> x.x.x5:443            Route   1      0          0
  -> x.x.x4:443            Route   1      0          0
TCP  x.x.x19:465 wlc
  -> x.x.x9:465            Route   1      0          0
  -> x.x.x8:465            Route   1      0          0
TCP  x.x.x19:993 wlc
  -> x.x.x9:993            Route   1      0          0
  -> x.x.x8:993            Route   1      0          0

Questions:
1) Should the backup server list connections its received?
2) If not, how do I verify that it's updated its internal tables?
3) Does it work if I always run both a master and a slave sync daemon at the 
same time on both directors, even if ipvsadm is only running on the master?

Regards,

Sebastiaan Veldhuisen










<Prev in Thread] Current Thread [Next in Thread>