I have a working Zabbix 1.8.2 server collecting data for our office and our colo facility. However the link between the colo and office is flaky. What I'm trying to do is setup a proxy on the colo side to have a 1 hour cache and relay the data to our primary server at the office. Our zabbix server is compiled from source and uses a mysql database
I've followed the instructions found in the zabbix documentation to compile the proxy using a sqlite3 database. I add the proxy to zabbix under Administration->DM->Proxies. The zabbix server "sees" the proxy because the "last seen" field is always under 60s. However when I assign a colo host to the proxy I stop receiving data from it. The colo host's zabbix_agentd.log file says this:
29343:20100622:124847 Timeout while answering request
29343:20100622:124847 Getting list of active checks failed. Will retry after 60 seconds
The zabbix_proxy.log says this.
2041:20100622:123131.760 Deleted 0 records from history [0.000994 seconds]
2028:20100622:124131.671 Error while receiving answer from server [ZBX_TCP_READ() failed
I also am unable to receive any SNMP data which is more important to me than the zabbix agent data. Has anyone had this problem before?
Zabbix Server OS: CentOS5.4 Zabbix Server Build: 1.8.2 from source Zabbix Proxy OS: CentOS5.4 Zabbix Proxy Build: 1.8.2 from source
P.S. The SQLite database on the zabbix proxy never gets any data written to it, it is identical to when I created it from the blank schema in zabbix-1.8.2/create/schema. (Yes I've checked the permissions)
I had this same issue a few weeks ago on a host.
After installing a new host with the zabbix-agent i got the "ZBX_TCP_READ()" in the Zabbix webpannel.
The solution was to change the Server value in the zabbix_agentd.conf to IP instead of HOSTNAME.
The problem originated from a nameserver defined in /etc/resolv.conf that was lagging for some reason.
I don't know if it's related to your problem, but it's worth a try.
if you are using active items, you must set proxy ip as the first entry in the agentd daemon config file, "Server" directive. if its passive checks, still, proxy ip must be in the "Server" directive so that connections from it are allowed.
snmp is a different thing, but in general i have two guesses :