As far as downtimes go, most suck. Then there are some that are more like happy little opportunities to learn something new. When Gitlab in my homelab started returning 500s, I had no idea how blessed I was. I was at best annoyed.

Looking at my git history, I've been happily running InfluxDB for years. However nearing the end of 2021, after long time of peaceful coexistence among my monitoring services, some more or less pressing problems emerged:

• The long-term future of InfluxDB 1.x is uncertain, it's pretty clear that I'll have to migrate to 2.x in foreseeable future.
• Memory usage becomes a bit problematic. Well, that's quite an understatement. The container ran out of memory frequently and I had to bump up the limit couple times already. Despite that, I can't even graph longer time spans without getting InfluxDB OOM killed and it's at this stage by far the most memory intensive service I ran. There are presumably some improvements in InfluxDB 2.x that could help here.

The solution seems pretty obvious, however in the end this post isn't really about InfluxDB.

In my homelab I use LXD/LXC quite a bit. Linux containers provide pretty decent alternative to full blown virtualization with smaller overhead and some extra perks like easier local filesystem access. All things considered, I’ve been happy with LXD for many years.

However there are places where the illusion of completely separated virtual machines shows some cracks when you look close enough. This is a story of one such crack that was haunting me for months.

This is an extension of very helpful article Running Docker on WSL2 without Docker Desktop (the right way) by Felipe Santos. My post attempts to go further and solve a very specific issue within WSL - the absence of nftables support. This is something you might not encounter unless you start running docker in docker or attempt to run docker images that rely on nftables support in some way.

I tend to use Molecule quite heavily and I often end up using docker to emulate actual full blown linux systems or kubernetes cluster in a container, which is where the absence of some kernel features shows up.

There’s a plenty of articles out there on VFIO GPU pass through out there, so I’ll skip many of those details here to let us focus on stuff that’s somewhat specific to Dell R711.

Before we get to the nitty gritty details of the whole setup, just a short warning: If the GPU pass through is the only use case you want to cover, starting with Dell R710 is absolutely not the way to do it.

Here’s why: