i am still completely new to the topic monitoring and hope to get some answers here.
1.) If I rely on Netdata as a solution, what exactly do I have to do to run the Netdata server offline (i.e. in my data center).
2.) We have some cloud solutions based on Tomcat with Apache and Java. On the webservers we installed on a JavaMelody, but nobody at our company really knows how to analyze these metrics. I assume that I will install a Netdata Agent on these Linux servers, which will deliver values to the Netdata server. Are they then in real time? As a Laye I can quickly see where the problem is. Or how do I best “learn” to monitor and interpret something correctly?
Oh, before I forget. There are several Tomcat instances running on the Linux server, i.e. various URLs end up on different Tomcat instances.
When you install Netdata, by default it will listen all IPs on the host and you can access it using http://IP:19999. The simplest way to install it is using our static installation.
Netdata has different collectors that will collect data from your web servers, you can take a look to configure them here and here.
Netdata monitors in real time the whole information, and you do not need to worry, this won’t be a pain for the computers.
Right now Netdata gives information on charts and you can see more details about them here.
Finally, let us know if you have additional doubts.
What speaks against an installation with the installation script?
If I have understood it correctly, I can install e.g. first on a dedicated Linux computer the first instance of netdata. There I get then the WebGUI on the port 19999 indicated.
Then I install the same script on the systems to be monitored, but change a few config files so that they send their data to my first instance (and thus not to the public netdata cloud).
I’m not an expert on this area in particular, but sounds like what you are talking about is maybe streaming to a netdata parent. The netdata parent can then sit wherever in your infra and then receive data streamed in from the child nodes.
So on your parent then you could go to /host/ for each child streaming into the parent via a url like below:
Yes, the streaming should fit, so I found a tutorial. But what I would still like to know. If I have e.g. a parent and 10 children and of my children 1 system has a problem, is it shown to me “immediately” on the dashboard of the parent, or do I only see it when I click in via “host”? that would not be practical.
When you Install using script netdata-installer.sh, it will compile direct on your machine and you will need to have some packages installed to have all features, thekickstart is static linked, so you won’t have these kind of concerns.
If I have understood it correctly, I can install e.g. first on a dedicated Linux computer the first instance of netdata. There I get then the WebGUI on the port 19999 indicated.
This is correct.
Then I install the same script on the systems to be monitored, but change a few config files so that they send their data to my first instance (and thus not to the public netdata cloud).
Yes, you will need to configure the stream.conf. You will have a network configured with a parent and many children.