Hi, Thanks to suggestions from the beta testers I've done more work on the Grafana setup. It's probably still not quite ready yet but could do with a bit more testing so I've opened it up to everyone now. So please if you have a few minutes can you take a look at: https://tools.bitfolk.com/grafana/ I've made it auth off of the Panel, so there is no need to enter credentials at multiple places any more. It should redirect you around as necessary. CPU graphs in Cacti at the moment show as % of a core, so if you have 2 vCPU (as all BitFolk customers do) then the graph can show as high as 200%. I copied that behaviour into Grafana but feedback was that people wanted it to be % of total CPU, i.e. strictly between 0 and 100%. So I did that. The traffic direction is also now presented from the perspective of your VM - Cacti shows from the perspective of the BitFolk bare metal host so the label for "In" means "Out of your VM". I thought it might be less confusing to flip those directions from the start for this. If you have multiple VMs then there will be a dashboard for each one. Click the icon on the left that looks like four squares and then "Manage" to see them all. I had feedback from one person that graphs with negative Y axis are less clear, but then the few other testers who got back to me on that said they don't see a problem with it. This was however before getting chance to compare and also it's still a really small sample of people. So I've added alternate dashboards that don't use a negative Y axis. I'd like some feedback on which of the two you prefer. Whichever of those is most preferred I will leave as default, but will probably still keep the alternate. And of course, any more feedback about how it looks and is laid out would be welcome. Also other feedback about what more could be done, though as I say initially I am only trying to replace and retire Cacti. Mini FAQ (will go into a wiki page at some point): Q. Will you be importing the 5+ years of data from Cacti? A. Sadly not. Prometheus doesn't have that sort of input feature (yet): https://github.com/prometheus/prometheus/issues/535 I will keep the data around and visible, but it won't be updated. If it later becomes possible to import then I'll try to do that. Q. How much retention will there be in Prometheus? A. The default setting is just 14 days, but I've optimistically set it to 1 year. Depending on how it goes that may need to be shortened or it might be fine. I understand that external [to Prometheus] storage can be used with something like Thanos keeping the older data while Prometheus only deals with the more recent: https://improbable.io/blog/thanos-prometheus-at-scale So when the time comes I'll look at switching to that sort of thing, but it doesn't need to be dealt with just yet. Q. Will you be allowing it to graph other things than the dashboards you've supplied? A. Maybe. The offer was always there for Cacti, but there is currently only one 1 customer making use of that. Conceivably if you know what you want graphed we can come up with a dashboard that shows it. This may require you to install an agent of some kind (e.g. node_exporter). Full edit access to BitFolk's Grafana is not likely to happen any time soon. Q. Will you expose Prometheus metrics related to our VMs so we can scrape them ourselves? A. Probably not any time soon. The current metrics mostly come straight from node_exporter on our bare metal hosts, they're mixed all together with other customers and infrastructure metrics. The access control would be a lot of work. Q. Can I share links from Grafana? A. You can explicitly share snapshots of whatever you are looking at. There's a "Share dashboard" icon on the top towards the right. It offers you the option of sharing a "Link", "Snapshot" or "Export". "Snapshot" is the one you want. The resulting link will let people see whatever you are seeing. It won't update. "Link" would show people an interactive dashboard, but its only available to authorised users, which basically means you and the admin account, so not very useful. "Export" is for getting the JSON definition of the dashboard out, and since you don't have access to the metrics is again not very useful for you. Cheers, Andy -- https://bitfolk.com/ -- No-nonsense VPS hosting