I have some ~6 years old never updated netdata installs working fine on my home servers, router etc. I understand those well, I modifiied some plugins, solved bugs myself, have written some new data collector plugins for my own things. Now I’m building a new (home) server and installed a fresh nightly NetData. And I don’t get it… Am I getting old or what happened? My first problems:
There is the settings icon on the top left, but it’s disabled. Why and how to enable it?
A lot of charts show grouped data sums. I can ungroup them via the Group by dropdown thing, but how can I make this to be permanent? Do I have to edit dashboard.js for that? For example Mount points → Utilization shows grouped sums of all mount points which is useless, says nothing to me. I want to see each mount points free and used space. But I have to ungroup this chart every time I want to see the data.
I clicked on the green + icon, to create a new dashboard, maybe I can create one which suits my needs. It says “Create a new Space”, okay, who cares how this is called, I entered a name, clicked Save and nothing happened.
And many more things like these.
What am I doing wrong? Do I need to sign up to the cloud services of NetData to make it usable?
First of all, great to see feedback from our long time users. A lot did change, but hopefully for the better. Some of the elements you see (and mentioned) are related to Netdata Cloud. However, the answer to “do I need to sign up for Netdata to be usable” should be “no”.
The UI that you are seeing was introduced about half a year ago to unify the interfaces of the local Agent dashboard and Netdata Cloud. They are now the same code base, and while some features are Cloud-only, the vast majority is available to the local dashboard. For multi-machine deployments where the Agents are in a child-parent configuration, this UI now also allows for seeing charts that cover that whole infrastructure, instead of node-by-node.
Some of the items you mentioned are artifacts of the optional relationship with Cloud. I’ll go through them one by one.
The settings behind the gear icon are for Cloud-only features, including managing the Cloud space, rooms, users and Cloud notification integrations. If the Agent is connected to Cloud, the user has logged in there, and the Agent was trusted, all of the Cloud features will work from the local UI as well.
The only non-Cloud setting currently is for changing the UI theme between light and dark, and that’s behind the user icon in the bottom left corner.
As the Cloud UI has always focused on observing your whole infrastructure, we believe compound charts are the best way to visualize metrics and help troubleshooting. However we do understand that some users would prefer to have separate charts instead of into a single one, grouped by a given property.
The solution we have in mind to address this, is so-called expanded views, that allow for drilling down on a given chart.
Until we have made these changes, we will most likely keep the previous UI available in the agent. You can find this at http://localhost:19999/v1/. Note, however, that this interface hasn’t been maintained in quite a while and you might hit bugs as a result of progress made on the Agent’s backend and collectors.
This is a bug and I just filed an issue for this. Thank you.
Naturally, we think that using Cloud provides a more complete experience, however, we’d love to hear about the many more things you mentioned. Can you expand on this?
Thank you for the detailed answer. Meanwhile I found the /v1 dashboard (read somewhere in the docs). Would be good, but the first thing I checked (and use most of the time) is Applications, but it is missing there.
The “many more things like these”:
Grouping/summing data which (I think) has no meaning. For example at the top of the page the “Average CPU per Node”. I assume, this is the average cpu usage of all nodes. What can I do with this? Lets say, it is at 25%, very nice. But 3 of my nodes are dead, because the services on them crashed and those 3 do nothing, and the 4th is working at 100%. That’s 25%. The same goes to Total Network in/outbound. Will it sum my router’s network i/o with the servers net io? Can not do anything with this data either.
And there are many like those above. For example lm-sensors temperatures only shows the current max, which will be one of the CPU core sensors. But this data is not enough, I want to see other temperature values too. For example I will never see that the motherboard temperature rises, as one of the CPU core temp sensors will always be higher than the motherboard temp.
Basically when I get alerted by NetData, I open up the dashboard to examine data. I don’t examine aggregated data, I’m looking for the problem, which can not be seen on those charts without ungrouping them.
Cloud. I don’t think it can add anything to my workflow, but even if it can I don’t want to stick to anything out of my control. I would like to use NetData without Internet connection too.
I see the concept, that the main charts can be seen on the Nodes tab for every node. I can click on one of the nodes name there, and I will be sent to the Metrics tab of that node. But at least the top charts on the metrics tab shows all the nodes together. I don’t get this. Those all node metrics (if somebody founds them useful) should be on the Nodes tab, and the ones showing one node only on the (Node) Metrics. Something like that.
I’m not an expert on the topic, so maybe I see things wrong. And I spent only 2-3 hours with the new Netdata, so I can be wrong with the above statements.
@X7UpoafA I whole heartedly agree with you on all points.
I can only add that ever since the change to group everything I am Using Netdata much less than before.
My use case was to be able at a glance to see if there are any issues, or to check the status of individual cgroups/services, disks, etc. This is simply not possible today unless I join the netdata.cloud and create custom dashboards. This is hardly easy and requires specific setup in advance to be useful, while still misses out on the ad-hoc possibilities the old way had.
I am not leaving netdata, but I do not always keep it running any more because, simply, it offers much less benefit and value than it used to.
It is also worth noting that the new UI is closed source/proprietary. This, IMHO, is also something that goes against many user’s principles.
I was a bit hasty. The main local UI is open, but it can also load proprietary code if the netdata server has Internet access. This can be disabled if cloud feature is disabled in cloud.conf.
My first problems: