-
Pandora Master Server connect to Agents?
Hi,
Is it possible for the Master Server to connect to the Agents, and not the other way around which seems to be the standard method. The reason i ask this is that the servers i want to monitor are not allowed to access the Master server (network restrictions). But it is possible for the Master Server to connect to the other servers.
BR,
Christian -
20 Replies
-
-
Hi,
Thanks for your fast answer.
I want to monitor for instance disc usage, memory usage, cpu usage primarily on Linux servers but also on some Windows servers.
Which of your suggested methods allows this?Are there some easy to follow guides that describes how to configure this for both Linux and Windows servers?
As far as I know the Agents send information to the Master Server so in my case they are unnecessary to use because this is not allowed in my network.
Have I understood this right?BR,
Christian -
Yes of course. With Windows is more easy you can use either SNMP or WMI, personally I prefer the second. Check the documentation and get a WMI explorer tool and/or a SNMP OID Browser to collect your WMI queries and/or your OID snmp for that. In module database coming with 2.1 you have several.
To monitor Linux remotely, use the SSH Server Plugin, I’ve write a guide on how to use, get it on:
http://pandorafms.org/index.php?sec=community&sec2=repository&lng=es&action=view_PUI&id_PUI=91
-
-
-
-
-
-
Thanks! Now the documents are readable!
I would like to clarify things in the documentation.
It assumes that an agent is already configured. Adding an agent is done under Administration -> Manage Agents -> Create agent.Also, the documentation says:
“Now go to your REMOTE SERVER admin mode screen, and create a new plugin module like this (using your own IP address for your REMOTE SERVER):”This means. In the menu go to Administration -> Manage Agents. You should have at least one agent in the list. Hover (put mouse pointer over) the agent’s name in the list with your mouse and you will get a menu. In the menu click on Modules.
I have successfully set up the AvgLoad module. But now I have another problem. How do i initialize the modules? I have 5 servers that i want to monitor with the AvgLoad module.
If i go to Operation menu -> View agents, i get this summary:
Monitor checks 5
Monitors normal –
Monitors warning –
Monitors critical –
Monitors unknown –
Monitors not init 5
Alerts defined –
Alerts fired –
Summary
Total agents 5
Uninitialized modules 100%BR,
Christian -
uccessfully set up the AvgLoad module. But now I have another problem. How do i initialize the modules? I have 5 servers that i want to monitor with the AvgLoad module.
If i go to Operation menu -> View agents, i get this summary:
Monitor checks 5
Monitors normal –
Monitors warning –
Monitors critical –
Monitors unknown –
Monitors not init 5
Alerts defined –
Alerts fired –
Summary
Total agents 5
Uninitialized modules 100%BR,
ChristianNon init modules are modules who cannot be executed by some reason, let’s review some details:
a. Have you checked the SSH key configuration and works from your pandora server, using root against your other servers, using user you want (make a manual ssh first as seen in docs)
b. If SSH works, command return data readable by pandora ?
c. Have you a plugin server running ?
d. Agent are using that plugin server ?
e. Have you checked all module parameters twice (destination IP address, user, command).
-
Non init modules are modules who cannot be executed by some reason, let’s review some details:
a. Have you checked the SSH key configuration and works from your pandora server, using root against your other servers, using user you want (make a manual ssh first as seen in docs)
It works.b. If SSH works, command return data readable by pandora ?
It works.
c. Have you a plugin server running ?
I don’t know. How can I see that?
d. Agent are using that plugin server ?
I don’t know. How can I see that?
e. Have you checked all module parameters twice (destination IP address, user, command).
I can’t find that fields. See attached file.
BR,
Christian
-
If this module it’s supposed to be a plugin module, WRONG :-), this is a data server module (alias local component). You need to define the module again AS a plugin module.
Non init modules are modules who cannot be executed by some reason, let’s review some details:
a. Have you checked the SSH key configuration and works from your pandora server, using root against your other servers, using user you want (make a manual ssh first as seen in docs)
It works.b. If SSH works, command return data readable by pandora ?
It works.
c. Have you a plugin server running ?
I don’t know. How can I see that?
d. Agent are using that plugin server ?
I don’t know. How can I see that?
e. Have you checked all module parameters twice (destination IP address, user, command).
I can’t find that fields. See attached file.
BR,
Christian -
-
-
AvgLoad is configured correctly now according to the guides. I have tried to run /usr/share/pandora/util/plugin/ssh_pandoraplugin.sh -h
-u root uptime | awk ‘{print $10}’ | tr -d “,”
and it returns a value like “0.07”.But there are no initialized modules. How to proceed?
BR,
Christian -
1. Your defined plugin module is defined as a generic_data type ?
2. Have assigned the agent to your server ?
3. Do you have a plugin server running ?.
4. The plugin server is running with the same user you have tested in command line ?If get no clues, use verbose 5 in pandora_server.conf and check the pandora_server.log and pandora_server.error in /var/log/pandora
Good luck and be patient, I’m sure your almost done ! 🙂
-
AvgLoad is working fine now! But there are still some uninitialized modules and all has to do with SNMP. I added the “Linux server” module group to my 5 servers. Do they require that an agent is running on the monitored server? (I can’t use agents because of network restrictions.) If I go to the SNMP console it says no SNMP traps in the database. If I go to Tactical View i see that Network and Plugin show 100%, the rest shows 0%.
How can I initialize the SNMP modules?
BR,
Christian -
Congratulations, think that use the plugin in Pandora is one of the most advanced topics, the rest of pandora is easier 🙂
About the SNMP monitoring, you need to separate the view of the SNMP polling and the SNMP traps. SNMP polling is to order Pandora to execute a snmpget command against a SNMP device, like a router or a switch (or even a computer with a snmp agent installed), this is a sincronous operation (each X seconds). On the contrary, receive a snmptrap is a asyncronous operation (this means, could happen or not happen in a million of years), commonly used to receive “alerts” coming from the device, like for example when a switch down a port of it’s fan is too hot.
To use SNMP polling monitoring, just add a SNMP module in pandora, creating a new network module. Most of SNMP items report data in incremental mode (generic_data_inc), this means that when you request a value, it reports the “global” ammount of information, por example if you request a SNMP device how much traffic have in a network adapter, they will reply a big number: the total bytes collected since device power on. So you need to substract the last known ammount of bytes from the current one, and divide by the seconds since the last known data, this will give you the bytes/second data you need. This operation is all managed with Pandora FMS using generic_data_inc.
Using SNMP Traps is totally different, you can receive traps from anyone, without configuring anything (except the SNMP console). When you receive a trap this is shown on the Snmp console.
You can define an alert, based on OID (the code who identify a trap, something like 3.4.1.1.4.5.24.2), agent IP or custom data (data who can be in the trap). You can also say Pandora to “copy” the trap information to a especial text module in the agent, if the agent is defined, this is called SNMP Trap forwarding. -
-
This needs to be done with SQL in the database extension, complex to do without SQL knowledge. Anyway, if you have software agent-based monitoring (installing an agent), and you delete all your agents, in the next data packet, they will create the agent again.
If not and you want to delete specific modules, you can try the massive delete tool, is not exactly you want but it’s some quick way to do the same.
