Unsupported item key

[heybails]

Why would I get an Unsupported item key for an item that works when I run both "zabbix_get" from the server and "zabbix_agentd" from the node? One of the items in question is configured as follows:

Name: Ceph Status-Degraded
Type: Zabbix agent (active)
Key: system.cephdegraded
Type of information: Numeric (unsigned)
Data type: Decimal
Update interval (in sec): 30
History storage period (in days): 7
Store value: As is
Show value: ceph degraded (custom value mapping)
Applications: Ceph
Enabled

UserParameter=system.cephdegraded,sudo cat /tmp/CEPH_DEGRADED.zbx

(tried without sudo as well and the CLI works fine, but the gui shows same error)

Restarted agent

zabbix_agentd -t system.cephdegraded
system.cephdegraded [t|0]

zabbix_get -s<IP address> -k "system.cephdegraded"
0

zabbix user can run the cat command on it's own if needed. I also set the debug level to 4 and saw nothing in the agent log file.

I've also changed the UnsafeUserParameter to 1 and tried it. Same result.

[glebs.ivanovskis]

Are other active checks working? If not, try to make something simple like agent.ping work as active.

[heybails]

Thanks for the reply. In fact, I have a few other custom made monitors that work as active checks and they are working perfectly.

[heybails]

Just to add, I set debuglevel=4 on the client and nothing shows up in the log. It's like it's not reading it at all because of the unsupported key which doesn't make any sense because the test commands work.

[glebs.ivanovskis]

Thanks for the reply. In fact, I have a few other custom made monitors that work as active checks and they are working perfectly.

On the same host and interface? Are you sure it's not a connectivity problem?

[heybails]

No, different hosts. But if it was a connectivity problem the test commands wouldn't work. I would not be able to run zabbix_get from the server to the node

[glebs.ivanovskis]

zabbix_get works like passive check. Server connects to agent. In active check agent must be able to connect to server. Try to connect with telnet from agent to server (default port is 10051) and make agent.ping or agent.version work as active check.

[heybails]

I fixed the issue. It's what I thought was the problem. These are openstack virtual servers that have a unique hostname that doesn't match the DNS name for the node. When I set the hostname to what is actually on the server, it started functioning. Now I have to figure out how to get all the openstack servers monitored. There are various clients that have the same hostname so I haven't a clue how to get around that.

Thanks for the help

[dmayorgap]

I fixed the issue. It's what I thought was the problem. These are openstack virtual servers that have a unique hostname that doesn't match the DNS name for the node. When I set the hostname to what is actually on the server, it started functioning. Now I have to figure out how to get all the openstack servers monitored. There are various clients that have the same hostname so I haven't a clue how to get around that.

Thanks for the help

Hi!, i have the same exact issue, you changed the host name field in the host to match the name of your virtual machine? Thanks!

[heybails]

I found that the best way to make these virtual hosts function smoothly in Zabbix is to build discovery rules and discovery actions. That way Zabbix finds it the way it should be laid out from the beginning, regarding hostnames, DNS entries, and IP's.