no data from aix hosts

[relikwie]

Hi I have installed precompiled AIX 5.1 binaries on two AIX 5.2 boxes, copied the the 3 binaries (zabbix_agent(agentd, sender) to /sbin. created zabbix user, folder for pid and log file, copied standard config from the zabbix server /etc/zabbix/zabbix_agend.conf over. Started with /sbin/zabbix_agentd and a ps shows multiple zabbix_agentd porcesses running on both nodes. But after adding these hosts to zabbix web, no data for at least an hour received.

Log files are sparse, no clue there. Anyone who does?


Thanks.

[James Wells]

Greetings,

Hi I have installed precompiled AIX 5.1 binaries on two AIX 5.2 boxes, copied the the 3 binaries (zabbix_agent(agentd, sender) to /sbin. created zabbix user, folder for pid and log file, copied standard config from the zabbix server /etc/zabbix/zabbix_agend.conf over.

Stupid question, but please bear with me. Did you start the zabbix_server daemon? Your write up implies that you are just running the agents.

[relikwie]

Hi James, a correct question, I wan't clear in my post.

The zabbix server runs on a other machine and succecfully monitoring a couple of windows servers and the zabbix server self. Now I wanted to add the aix machines for monitoring. So downloaded the agents and installed them like described in my post. /etc/zabbix/zabbix_agend.conf on the aix nodes do point to the server (a debian box).

could it be that I used agents compiled on AIX 5.1 and I have 5.2 running?

Regards,

[Villain]

make sure you edited the agentd.conf file to hold the correct names of the computers you're working on. That was the first mistake I made when copying the 'example' config over. I forgot to go in and make the necessary changes for each computer to be identified.

[James Wells]

could it be that I used agents compiled on AIX 5.1 and I have 5.2 running?

Possible, but unlikely.

Try this, from the server, telnet to the Agent port, 10050 by default. If you get a connection, you will see the usual telnet connection established which means that the agent is configured to accept connections from the server. If not, then your agents are not allowing connections from the server. This is controlled by the "Server" entry in the zabbix_agentd.conf file.

If you are connected type 'agent.version'. This will tell you if the version of the agent is at least partially operational. If it fails, then you are most likely having an issue with the version of the Agent.

If both of these work, create an item of type Zabbix Agent, that checks agent.version, with a delay of 20 seconds and stores the data as log. Let this run for a few minutes and check to see if you are getting data. The key here is that you use type 'Zabbix Agent' and not 'Zabbix Agent (Active)'.

[relikwie]

Oke, seems that telnet is not working. And if I look on the aix boxes I notice that the agent isn't listening at all.

On the aix node:

/etc/zabbix/zabbix_agentd.conf

Server=10.32.32.201
Hostname=zabbix.domain.com
StartAgents=5

processes (forks as configured in conf)

root /home/root/ # ps -ef | grep zabb
zabbix 40886 82948 0 21:43:52 - 0:00 /sbin/zabbix_agentd
root 45724 86374 0 22:00:15 pts/7 0:00 grep zabb
zabbix 67376 82948 0 21:43:52 - 0:00 /sbin/zabbix_agentd
zabbix 70580 82948 0 21:43:52 - 0:00 /sbin/zabbix_agentd
zabbix 73670 82948 0 21:43:52 - 0:00 /sbin/zabbix_agentd
zabbix 79682 82948 0 21:43:52 - 0:00 /sbin/zabbix_agentd
zabbix 82948 1 0 21:43:52 - 0:00 /sbin/zabbix_agentd

netstat -an

tcp4 0 0 *.13 *.* LISTEN
tcp 0 0 *.21 *.* LISTEN
tcp 0 0 *.23 *.* LISTEN
tcp4 0 0 *.37 *.* LISTEN
tcp 0 0 *.512 *.* LISTEN
tcp 0 0 *.513 *.* LISTEN
tcp 0 0 *.514 *.* LISTEN
tcp4 0 0 *.515 *.* LISTEN
tcp4 0 0 *.543 *.* LISTEN
tcp4 0 0 *.544 *.* LISTEN
tcp4 0 0 *.631 *.* LISTEN
tcp4 0 0 *.657 *.* LISTEN
tcp4 0 0 *.1334 *.* LISTEN
tcp4 0 0 *.9090 *.* LISTEN
tcp4 0 0 *.10000 *.* LISTEN
tcp4 0 0 *.62221 *.* LISTEN
udp4 0 0 *.13 *.*
udp4 0 0 *.37 *.*
udp4 0 0 *.518 *.*
udp4 0 0 *.631 *.*
udp4 0 0 *.657 *.*
udp4 0 0 *.48085 *.*
udp4 0 0 *.48091 *.*
udp4 0 0 *.48092 *.*
udp4 0 0 *.48093 *.*

Here the telnet from the zabbix server:

zabbix:~# telnet aix01 10050
Trying 10.32.32.196...
telnet: Unable to connect to remote host: Connection refused

Regards,

[Villain]

if the aix machine you are trying to get information from is named zabbix.domain.com, then you are fine. However, if it is named aix01 (as your telnet session seems to indicate), then the hostname should be simply aix01.

the hostname parameter is poorly defined. It should be listed as a ClientID or even AgentID, which would make it less confusing. This entry should be the name you configure in Zabbix under 'Configuration' -> 'Hosts'.

For example:

Server = zabbix.mydomain.com

Web Server = web.mydomain.com

In my agentd.conf file, I have:

Server=10.1.1.10 # zabbix server address
Hostname=web # unique name for this server
#ListenPort=10050 # uses this by default, even if commented out
#ListenIP=127.0.0.1 # left commented out to use all IP Addresses

I hope that this helps sort the issue out.

[relikwie]

Villian, about your comment before.
All there is to change in the agentd conf is the ip address of the zabbix server, am I correct. That was the case for the windows clients, a copy of one config containing the zabbix server ip so it is allowed to connect.


Thanks,

[Villain]

Villian, about your comment before.
All there is to change in the agentd conf is the ip address of the zabbix server, am I correct. That was the case for the windows clients, a copy of one config containing the zabbix server ip so it is allowed to connect.


Thanks,

Actually, the hostname parameter is used to uniquely identify each computer. This should be different on each agent computer.

[James Wells]

if the aix machine you are trying to get information from is named zabbix.domain.com, then you are fine. However, if it is named aix01 (as your telnet session seems to indicate), then the hostname should be simply aix01.

Actually that is only needed if you are using Zabbix Agent (Active) items. In his case, he is not even getting to the point there that is a concern.

Here the telnet from the zabbix server:

zabbix:~# telnet aix01 10050
Trying 10.32.32.196...
telnet: Unable to connect to remote host: Connection refused

Okay, so it is not listening to port 10050. Check your zabbix_agentd.conf file for the ListenPort parameter. By default this is set to 10050, though I noticed that you are showing something listening on 10000, which is the old port that the Zabbix Agents listened on, in version 1.0. If your ListenPort is configured to 10000, then try the test again, against the correct port.

Actually, the hostname parameter is used to uniquely identify each computer. This should be different on each agent computer

If you are using Zabbix Agent (Active) mode, this is needed, but it can also be bypassed, if your actual hostname is set the same as the entry in the DB. IIRC, there is code in the agent that allows you to bypass this entry completely.

[relikwie]

Villian, yes, I wrongly put the zabbix server's name in the "Hostname" field. It's indeed poorly pointed out (not clear at all). But I changed it with the actual servername, FQDN and just the name "aix01". Each time I tried I also restarted the agent. But with no success.

I'am afraid I have a depper problem here. pity the agent is very sparse at logging. I have set it to level 4 and all it does is logging start and stop events.

Thank you guys.
One thing is for sure, people around zabbix are very helpfull and kind

[relikwie]

James, bingo. Sharp. I feel its the 10000 port, in case ZBX stands for zabbix.
See below:

zabbix:~# telnet aix01 10000
Trying 10.32.32.196...
Connected to aix01.<domain>.com.
Escape character is '^]'.
agent.version
ZBX_NOTSUPPORTED
Connection closed by foreign host.

So, the binaries available for download here http://www.zabbix.com/download2.php are not useable? I'll try to compile from source tomorrow and see it that helps.

Thanks.

[Villain]

true, if the dns name isn't there and the hostname matches in the db, you might be able to get away without the hostname parameter. However, when we did this (and we could ping by the same hostname), it didn't work. So, to be safe, we started ensuring the hostname matched the db entry. Since then, we haven't had any problems doing active or passive checks.

As for this issue, are there any firewall rules that would be blocking these connections? If not, try the ListenPort=10050 and ListenIP=10.x.x.x entries in the config file. That should cover it. Make sure you kill the agentd and restart it, then try the telnet session once more.

[James Wells]

So, the binaries available for download here http://www.zabbix.com/download2.php are not useable? I'll try to compile from source tomorrow and see it that helps.

They work, however, you need to be using the Zabbix 1.0 server to use them. Those binaries are for Zabbix 1.0, instead of Zabbix 1.1, which does not support the hostname at all, nor does it support Zabbix Agent (Active) mode.

[relikwie]

As for this issue, are there any firewall rules that would be blocking these connections? If not, try the ListenPort=10050 and ListenIP=10.x.x.x entries in the config file. That should cover it. Make sure you kill the agentd and restart it, then try the telnet session once more.

Done it as you describe and as suspected by james, default it ran on port 10000. Now I can telnet on 10050 with the same zbx_unsupported on agent.version as a result. And also server still does not receive data.

Regards,