z-push-top not working with cluster
-
Hi, we build a cluster with a layer 7 haproxy load balancer and two z-push servers with mysql backend and memcached ipc.
if we use z-push-admin it works, instead z-push-top does not work, it does not show any connection, it’s empty.
z-push 2.5.0 in both backend.
Any ideas? Thank you -
Hi @frayber,
Which backend exactly are you using?
Does z-push-top start without errors?
Is by any chance TOPCOLLECTOR_DISABLED set to true?
Are there any errors in z-push.log?
Did you checked if there’s actual data in memcached? Where is memcached server installed? On the load balancer? On one of the Z-Push servers?
Manfred
-
Im’ using z-push zimbra backend.
z-push-top starts without errors.
TOPCOLLECTOR_DISABLED it isn’t defined.
No error related to z-push-top in z-push.log
I have two memcached servers, installed on both backends, configuration is:define(‘MEMCACHED_SERVERS’,‘vm-lxzpush1:11211,vm-lxzpush2:11211’);
memcached seems populated, with about 90% of hits, but with low usage (1.8%) on total 64 MB (32 each server) and 60% wasted, are they normal values?
Thank you -
Now I see 27 open connections with red status, with very big time, not changing; yesterday I saw 14 open connections.
But they are over 200, without cluster I saw all them. -
I think tha cause of problem is memcache configuration, because if I use only one memcached server on first backend:
define(‘MEMCACHED_SERVERS’,‘vm-lxzpush1:11211’);
it works good.
Any suggestions? Maybe shoud I use memcache.session_redundancy and memcache.allow_failover ? -
Bug opened: https://jira.z-hub.io/browse/ZP-1523
