[lvs-users] probs with sphinx/searchd

murray lotnicz murray9000 at gmail.com
Tue Jan 13 19:29:54 GMT 2009


Hi all,

i've got 4 sphinx nodes [0.9.9-rc1] running behind a
keepalived/heartbeat LVS cluster
[centos 5]. load balancing was fine for a few months but now i'm
having intermittent
issues reaching the sphinx nodes via load balancer. i can telnet 3312
directly to each
node but not to the VIP. at first i thought it was an LVS issue but
this keepalived
cluster is also load balancing a couple other pools [apache + tomcat]
without probs. all
3 pools are DR [direct routing] with least-conn predictors. loopback
ints are good and
appropriate sysctl.conf hacks to prevent ARPing are verified. iptables
is not currently
running.

is anyone else seeing issues with LVS and Sphinx?? any input would be
appreciated, thanks.

murr




More information about the lvs-users mailing list