Traffic between Linux-Agent and Backend

Hi all, has anyone ever checked how much traffic happens in say an hour between a checkmk site and a Linux standard agent? There are no official statements/docs about this, are there?

regards
Christian

Hallo,
what is the problem behind your question?
Try to contact your agent from the server with telnet IP 6556 to check the size of datas for one transfer as a first step.
Ralf

2 Likes

There cannot be any official docs as the amount of agent data depends on agent plugins installed and e.g. number of filesystems, network interfaces or databases.

3 Likes

Hi @rprengel
we plan to use google cloud and we have to pay the inbound/outbound traffic. The specific question is how much data is generated between a standard Linux system in the cloud and our CMK backend. The idea with the simple telnet is definitely an approach. I will test this with one of our systems. Thanks.

You might even be able to build a check that runs regularly and reports the amount of data. That way you could measure your whole current landscape and give more than an approximation.

1 Like

Hi Chris,
try “cmk -d | wc -c” against a normal Linux agent to find out the size of bytes in data.
It should be ~50k - 100k bytes per Agent .
Greets,
Christian

1 Like

Would it be not better to run a monitoring instance also inside the cloud?
I have such a installation inside Azure and monitor all virtual machines there from the Azure instance. Outside connection is then only the status data or mail traffic if needed.

5 Likes

@ChristianM thanks for this idea!

@andreas-doehler , that is our plan as well. We are just collecting information that could be interesting for this whole topic, hence the question with the traffic. But we are already assuming that we will put our own site in the cloud.

Thanks for all the info, that was helpful! :slight_smile:

Just to provide some numbers.

We have ~ 750 hosts in our monitoring and ~ 50 Mbit/s incoming Traffic on the monitoring server’s interface, that’s ~ 67 Kbit/s per host (includes checkmk agents, SNMP hosts and ping only hosts).

Just to really throw some confusion into the mix, I’ve got a site with a little over 750 hosts (a little under 30k services, how about you?) pulling an average of only 8 Mbit/s :smiley:

It’s going to vary a lot by what your environment looks like. Some service plugins return LOADS of details, simple systems won’t have much to say. You can always cut the data rate in half by doubling your check interval too :stuck_out_tongue:

But yes, I would also advise running the monitoring site in the same… zone(? whatever the Google Cloud term is) so you’re not paying dollars on the monitoring traffic. If you’ve gotta pay for traffic between zones, go for another distributed site in the extra zones. The enterprise license lets you scatter sites pretty freely, it’s just the total amount of monitored stuff you pay for, regardless of the deployment structure.

Wish I could give you an exact figure on what the traffic of interacting with a site actually costs you, but I don’t have an interface monitored where that’s the only thing. The daytime traffic looks to be… maybe 300kbit/s above baseline average, if I had to pick a high estimate? Peanuts.

3 Likes

In addition to the hints already given, look into the directory ~/tmp/check_mk/cache. Here you can find the outputs of the last agent run for every monitored host.

If you stick to the default check interval of 60 seconds, the content of this directory denotes the traffic in a minute.

3 Likes

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed. Contact an admin if you think this should be re-opened.