I'm about to create a high available load balancer based upon heartbeat,
ldirectord and LVS with direct routing. Everything should be fine, the setup
is working, connections are balanced through my webservers. However, after
some time some ressource limits on the load balancer seem to exceed, since
I'm not able to getting a connection anymore.
But I don't see anything in the logs, nor any change. Even ipvsadm doesn't
show me any problem. Some details of the setup (somewhat shortened, however
not changed anyway):
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP 10.0.0.16:80 lc persistent 7200
-> 10.0.0.104:80 Route 1 677 518
-> 10.0.0.103:80 Route 1 678 597
-> 10.0.0.102:80 Route 1 652 547
ipvsadm -Lnc | wc -l
I enabled IP forwarding and added a virtual (shared) IP, managed by
heartbeat. However heartbeat is working well and not doing any takeover or
ressource give up, so I don't see the problem there.
At this point I'm able to connect to 10.0.0.16:80 and getting a working HTTP
response from one of the node servers. But after some time, without noticing
any change, I'm just getting "Connection refused", but nothing changed. Even
the output of ipvsadm -Ln still shows me my setup, with all nodes activated.
Problem solves as soon, as I reboot, restart ldirectord/heartbeat or such.
There might be a ressource problem I guess, maybe open file descriptors oder
such. But I can't figure out, what the problem could be.
The machine is x86 based with unchanged ipvsadm/ldirectord/heartbeat
packages from Debian Etch. It's not running out of memory nor having a
Email.it, the professional e-mail, gratis per te: http://www.email.it/f
Rivoluzione nei Prestiti: Finatel ti offre il superfinanziamento a Zero
Spese. Anche protestati e cattivi pagatori! Richiedilo
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=7369&d=20071223