[lvs-users] Orphaned checks from no longer active ipvs nodes
timo.schoeler at riscworks.net
Fri Nov 5 14:16:41 GMT 2010
-----BEGIN PGP SIGNED MESSAGE-----
I have a four node cluster running (CentOS 5.5 x86_64, heartbeat 2.1.4)
using ldirectord for ipvs stuff.
Imagine, I have the active load balancer running on node a (there's node
a to d, e.g.). All of those machines were rebooted recently.
In this state, everything is okay, and entities being monitored (like
real server 10.2.3.4 in the line below) only see checks from node a.
Now imagine, I move the resource over to node c. From now on, I'll see
checks from both, node a *and* node c, although it should be sufficient
to run those checks from node c.
(On node a:)
ps auxwww |grep tcp
root 7288 0.0 2.1 104272 11392 ? S Oct11 0:31
tcp:10.0.2.3:9000 checking 10.2.3.4
I can reproduce this behaviour. Is this intended or is it really
'orphaned checks running', as I might assume?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)
-----END PGP SIGNATURE-----
More information about the lvs-users