I miss the days when PG articles were actually worth reading, and not just spurious drivel from yet another person that hasn't been in touch with reality for decades.
Using Prometheus in federated mode has been fine for me, I'm running it almost exactly as depicted in the article and I'm not sure why OP describes it as a complex on - you just need to have some idea of what metrics you'd like to store longer term.
Similar boat here, and know of many other companies in the same situation (although we chose Alma).
We thought we were on a good proven path of paying for support where we need it, and using the downstream alternatives when we didn't. Turns out we should have more seriously considered a move to Debian...
I'm not really shocked Red Hat are trying to squeeze more money out of their existing audience, everything they offer is eye-wateringly expensive nowadays. I almost choked when I got a quote back for the Ansible automation platform, before installing AWX for free.
Another irritating move from Red Hat, likely driven by their unhappiness with the popularity of Alma/Rocky instead of a mass migration to RHEL they had probably hoped for.
As the source is still available, but not publicly, could it be that these alternate flavours enter in to a license agreement with Red Hat that's community funded?
Unfortunately the developers of those particular things (often like ad developers) use obscurity of things like div names to make the task less trivial.
Previously we were using FortiManager to manage a cluster which has CSV export functionality, however for reasons too long to explain on HN, we no longer use FM and simply use FG directly. We occasionally provide policy and DNAT exports to customers and internal teams, so this script replaces what we used to have and allows for a bit of on-the-fly editing too.
> So yeah, try feeling grounded with that. It's a miracle that it even works.
Just because something is too complicated for you to understand, doesn't mean it's no good. Is this guy going to decline machines/life support in hospital because he thinks he could build something simpler?
I self host some things, mainly websites, but trying to self host the amount of stuff in that article is a fools errand. You'll spend more time updating, patching, and failing to notice any gaping security holes. I'm quite happy to pay Apple about 79 pence a month to deal with all that for me.
I've been undertaking a similar project, using an ESP32 currently for testing, also have an Arduino Micro Pro to test yet, during lockdown/furlough. I'm hoping to customise and sell them on Etsy, I've sold code + 3D prints before, but I'm just taking time to get the PCB right before I get my first ever board printed. Aiming for hot-swappable switches and some LED magic.
Thanks, you've described what I had in my head as a rough way of doing things - the issue with this approach is that it appears to be one bot instance per user. I.e. if I want the same 'user' to be able to be invited to unlimited servers, that 'user' must run from a single application instance - meaning all server chats flow through a single app instance.
Yeah I have some projects at work which manage containers in this kind of fashion, which is probably why I'm applying that to this and over-engineering - which I almost always do.
The current bot could work with just passing in a token, and then fetching its settings from a REST API (for the web bit) instead of just a JSON file - my only issue with this approach is I think it'd require the user to setup a developer app with discord user, instead of the nice seamless 'invite and it works' approach other bots seem to have.
I'm still using a 2015 MacBook Pro that I'm clinging to, despite the screen having developed an annoying flicker every few minutes. I know 4 others in my office with Apple laptops, and they're all 2015 MBPs because until now, nothing else has been acceptable.
This seems to be like an answer to the "Can we just have the 2015 MBP with updated hardware?" question. Pending benchmarks, this could be the light at the end of the tunnel for a lot of users that don't need more than 16GB RAM (I'm still scraping by with Docker on my 8GB Mac and a lot of SWAP).
Chrome on Mac, it's convenient, and despite robbing all the RAM my 2015 MBP has to offer, it works well.
I tried switching to FireFox for two weeks, but found a lot of visual tearing/glitches, and it didn't have pinch to zoom which I use on Chrome constantly.
Not the answer you're looking for, but I found it too busy, unintuitive, and the lack of non-distribution for the OSS version finished off my test run. I use Graphite, have done for years, and am quite happy with it.
On a further note, I attended Grafanacon and saw a few keynotes given by the CEO of InfluxDB and have never been more turned off of using a product by a single person.