Zabbix agent unreachable

[Plaice]

Hi all,

We've recently upgraded our Zabbix server from 1.8 to 2.0.

Our old installation ran fine however with the new version we're seeing quite a few emails stating that agents are unreachable.

When this happens we're checking the servers and finding that the alerts are false and the servers are still up and responding.

The queue of items on the server isn't massive and we've even upgraded the VM with more CPU and RAM but are still receiving the alerts.

Anyone have any ideas of where to look first?

Thanks

[vintagegamingsystems]

Look in the logs

You should look in the log of the zabbix_agent.log on the hosts that are unreachable. See if there is anything weird happening in there.

Are any of your hosts reachable? Or just some?

Sounds like there could be a configuration problem in your zabbix_agentd.conf file. You should compare your old configuration files with your new ones.

You also need to provide more information about what is happening and where.

[Plaice]

We have 97 monitored hosts. The alerts we're receiving are seemingly random, not one specific host or group of hosts.

[Plaice]

Also, when we get the alerts they are only for one or two hosts not all 97.

[vintagegamingsystems]

Not very helpful

Your response is not very helpful in diagnosing your problem. Look at the logs for the agents in question.

[tchjts1]

Check to see what trigger is being used for that alert. There was a change from 1.8 to 2.x

The trigger that is used now for Zabbix agent unreachable is

{Template OS Windows:agent.ping.nodata(5m)}=1

[tchjts1]

Additionally, if you look on your Zabbix server in zabbix_server.log and you see a lot of "failed connection. retry in 15seconds" then try increasing your Timeout= value in zabbix_server.conf from 3 to 10, then restarting your Zabbix server process.

If you change that Timeout= line, you either have to remove the comment (#) before the line, or make a new line without the comment.

[Plaice]

Just checked the last couple of agents to throw up this alert and don't seem to be able to locate a log file. Checked the .conf file and log file location is pointing to C:\Program Files\Zabbix Agent\Zabbix_agentd.log however there is nothing there.

Have checked the trigger and it is indeed using that value.

[tchjts1]

C:\Program Files\Zabbix Agent\Zabbix_agentd.log

Zabbix is case sensitive, even on Windows. Does that path exist with that exact case usage?

[Plaice]

Location does exist, log file doesn't.

[tchjts1]

That's interesting. I think they changed it so the agent won't even start if it cannot write to the log. Are you looking at the correct conf file - zabbix_agentd.conf

Can you post one of your zabbix_agentd.conf files? You can remove any IP or server names if you like.

[Plaice]

Copy of zabbix_agentd.conf should be attached.

[tchjts1]

I'm not sure what to tell you on this one. Try going to that directory and creating the file "Zabbix_agentd.log". (I am not a fan of using any capital letters, though)

Something else I noticed in your zabbix_agentd.conf file is that you have
ServerActive=127.0.0.1

This means that if you are using any Zabbix agent (Active) items, your hosts are going to try to communicate to itself. That value needs to be either the IP or DNS of your Zabbix server.

Another thing I see is that for Hostname= you actually have "HOSTNAME" in that field, which is wrong... unless those 2 values above are what you put there in order to share this file.

[vintagegamingsystems]

A few suggestions.

Sounds like your agent may have never been started. But first make some changes to your configuration file as it has problems.

Make sure your hostname in your configuration file is the hostname of the host you are monitoring. Use the real name of host NOT the visible name.

If the visible name was test and the hostname was test.test.com. Use the latter.

Hostname=test.test.com

You can also put your ip for the Hostname, if DNS has failed.

You might try to change your log file to be in the root. Just put C:\ .
Something like this:
LogFile=C:\

Also make sure that you have installed the zabbix_agent onto the machine.

The directory that has the zabbix_agent software is something like ../bin/win64/zabbix_agentd.exe

You can use win32 if you have a 32-bit Windows OS.

You can install it by using the command.

../win64/zabbix_agentd.exe --config <configuration file> --install

If you put the configuration file in your C:\ then use a command like this.

zabbix_agentd.exe --config c:\zabbix_agentd.conf --install

Once it says that it is successfully installed run this command.

zabbix_agentd.exe --start

It should say it is successfully started.

If you have problems, take a look at this page.

https://www.zabbix.com/documentation/2.0/manual/appendix/install/windows_agent

Also, DNS may not be resolving you server name correctly, so try to use the IP address for the server:
Here is an example:

Server=232.234.233.12

You can find the IP address of the zabbix server by using the command:

$ host zabbix.createitg.com

[vintagegamingsystems]

Is port open?

Make sure that port that the zabbix_server listens on is open on your machine. Usually port 10050. Open up your firewall on your host machine for port 10050. It looks like yours might be 10053??? You should try to lock down that port once you get a connection, so that only the zabbix_server can listen to it.

For a test you could turn off your firewall temporarily and see if the agent starts to work.

Also check the version of the agent and see how old it is. The version I am running is 2.0.6.

You can check the version by using the command zabbix_agentd.exe -V or maybe -v