The Netdata agent has supported dark mode for a long time, but the Netdata Cloud, being only recently released, has not. Pictures are worth a thousand words, but keep in mind that this is still being worked on!
A lot of people have been asking about this and we now have a comprehensive answer.
I can use this thread as an excuse to get a second monitor…
The only part of my setup that is really interesting is a keyboard tray I built a few years ago.
Not a real butcher block, but a thinner panel with epoxy top and a rubberized underside. Mounted to desk using industrial drawer hardware and metal brackets.
Also pictured, a '93 IBM (Lexmark) Model M.
Because of the distributed nature of Netdata, before this feature, notifications about triggered alarms had to be configured and dispatched by the individual nodes, by the Netdata agent registering a triggered alarm. This becomes inconvenient in certain common situations - for example, when the node is down. With this new system in place, the Netdata Cloud will be able to send email notifications directly. Of course, we will also enable configuring and sending notifications for other alarms (not just for downed nodes). For some users, especially those monitoring larger infrastructures, this approach to notifications will be easier and faster than configuring each individual agent.
We also plan to add new notification integrations through web hooks, which will allow alerting through other platforms, like Twilio, Slack, or PagerDuty - and many more. We are also planning to ship a mobile app (iOS & Android) to let users receive notifications.
We have carefully documented all the steps and details to be able to build and package Netdata. This should be quite useful for package maintainers.
Looks like a bug, will track here https://github.com/netdata/netdata/issues/9635
It is time consuming to go over thousands of metrics and hundreds of charts to manually identify metrics which are correlated.
For a specific time window when an anomaly occurred it should be possible to automatically find a subset of charts where for the same time window the same anomaly was detected.
Let’s say a software issue resulted in high CPU and disk load over thirty seconds during an outage. Selecting that time window for the CPU load chart will provide users with the option to ask Netdata to find correlated charts. Netdata can then display the disk load chart automatically, while also hiding charts that are not relevant.
This will allow you to very quickly narrow down all the metrics being affected by a specific issue, saving a lot of time in debugging specific software or infrastructure issues.
We also want users to be able to share specific correlations they have found by sharing a link with their team.
A detailed tutorial about running Netdata on production systems.
Great video for SREs working with Netdata
You may have to create networks and assign hostnames for containers within those networks using
docker network create and following up with
docker run -h human_readable_hostname --ip [an ip on the network] --net [the network created]. These commands will have some additional options based on what exactly you are trying to run…
It might be easier to use docker-compose for something like that. In any case I think the trick is to set a readable hostname for the container.
We also have a lengthy docker guide but I don’t think it covers this specific question.
Right now it is possible to lock down a particular machine to only connect to Netdata Cloud externally. As per the ACLK guide for the moment we use a single IP address.
This is likely to change in the future, however.