2 ms·
There's a lot going on in this article, from a lot of different ends. One of the big tensions (among many) that I'll focus on is rootless containers. Frankly i
by rektide 10y ago
There's a lot going on in this article, from a lot of different ends.
One of the big tensions (among many) that I'll focus on is rootless containers. Frankly it's a bit silly to think containers are useful when you need root on every system to use them- do people on academic systems simply not get the ability to use containers? As we move towards multi-tenant, "hyper-converged" systems (things running lots of workloads), figuring out who/what to bless as OK is a critical bottleneck, one that needs a lot of trust.
So the idea has been gesticulating around of "rootless" containers[1]. There's a lot of recent excitement and work- @lordcyphar has started a dedicated website for the idea, https://rootless.containe.rs https://rootless.containe.rs [2], which gives a broad status overview of the pieces. The "runc" alternative to docker merged rootless recently, after a year in progress[3], and recent work focuses- as the article hints at- around the filesystem side of things. He's building orca to build oci standard images, and although strongly hinted[4] I'm not sure how exactly that factors in to supporting rootless.
[1] https://www.cyphar.com/blog/post/rootless-containers-with-runc https://www.cyphar.com/blog/post/rootless-containers-with-ru...
[2] https://rootlesscontaine.rs/ https://rootlesscontaine.rs/
[3] https://github.com/opencontainers/runc/pull/774 https://github.com/opencontainers/runc/pull/774
[4] https://twitter.com/lordcyphar/status/849951277744181248 https://twitter.com/lordcyphar/status/849951277744181248