Ohh! Amazing, I'll definitively give this a shot. I'm building docker images with nix2container and was working on .mstack compatibility - so I can benefit from nix layering.
This kinda fit's really well into some other systemd-nspawn endeavors I was looking into, single big VM/computer running systemd-nspawn containers - but I also just want to run docker as well for some services, so this fits perfectly into that gap.
Really much looking forward into the future compatibility of nested systemd-nspawn, this would give the entire setup an even higher value.
It answers in the shapes of API 1.51, and of the versions down to 1.24 that clients ask for.
The article starting in a really weird way (as if the introduction is missing) was suspicious, but this makes it pretty clear this wasn't written by a human; or at least a very lazy one.
Makes it incredibly hard to read, because why should I care at all? If it's worth releasing it's worth writing a couple sentences about each feature surely
That's exciting. I was a fan of the rkt idea and this goes in the same direction - we already have a process supervisor aware of namespaces. When trying to write my own version of that I ran into lots of issues with systemd's import just not supporting lots of cases for OCI and gave up. I'm glad they had a similar idea and succeeded.
pl | a day ago
Ohh! Amazing, I'll definitively give this a shot. I'm building docker images with
nix2containerand was working on.mstackcompatibility - so I can benefit from nix layering.This kinda fit's really well into some other
systemd-nspawnendeavors I was looking into, single big VM/computer runningsystemd-nspawncontainers - but I also just want to run docker as well for some services, so this fits perfectly into that gap.Really much looking forward into the future compatibility of nested
systemd-nspawn, this would give the entire setup an even higher value.I think what I'm going to investigate next is how
skoepoandmstackcan be combined - any ideas on that front? Issue was closed as not planned, but some pointer was left behind.yorickpeterse | a day ago
The article starting in a really weird way (as if the introduction is missing) was suspicious, but this makes it pretty clear this wasn't written by a human; or at least a very lazy one.
WilhelmVonWeiner | 14 hours ago
Makes it incredibly hard to read, because why should I care at all? If it's worth releasing it's worth writing a couple sentences about each feature surely
wofo | 23 hours ago
Looks like pangram agrees.
viraptor | a day ago
That's exciting. I was a fan of the rkt idea and this goes in the same direction - we already have a process supervisor aware of namespaces. When trying to write my own version of that I ran into lots of issues with systemd's import just not supporting lots of cases for OCI and gave up. I'm glad they had a similar idea and succeeded.