[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Podman 3.0 and Debian bullseye



On Mon, Jan 25, 2021 at 06:47:25AM -0500, Reinhard Tartler wrote:
> On the tradeoff "podman 3.0 with docker-compose" support vs.
> a "nomad driver for podman", I see more value for (more of)
> our users for the newer podman. I base that on popcon numbers:
> 
>  - nomand: 35
>  - nomad-driver-podman: 4
>  - podman: 340

I realize the thread has continued down that path, but I think that's a
false dichotomy. Debian sid has nomad-driver-podman 0.1.0-2, but
upstream has released v0.2.0, very explicitly deprecating the Varlink
interface in favor of support for the HTTP API¹.

nomad-driver-podman v0.2.0 may break Dmitry's use case (#69²), but
clearly this isn't a use case that upstream considered as important
enough, or one they considered release critical for that matter, as the
task was filed well before the 0.2.0 release. They have made no
commitments to fix this issue to date.

Overall, I think the choice here is between:

1) staying with a now-outdated version of nomad-driver-podman, until
upstream fixes a bug they consider unimportant, with no fix in sight. In
turn this requires podman's 1.x API, which is only offered in podman 2.x
as a backwards compatibility layer, and fully removed in 3.x.

2) upgrading both nomad-driver-podman and podman to their latest
upstream releases. These are seemingly compatible with each other, but
breaking one particular use case, which while niche, happens to be
unfortunately the maintainer's primary use case.

IMHO option (2) is the reasonable choice here, and one that benefits the
most users across the board. It sucks for Dmitry, I get that; but I
don't see how option (1) would work. It means that we'll get stuck on an
older stack, with no end date in sight.

The fact that as has been mentioned in this thread a) bullseye is around
the corner b) nomad-driver-podman isn't even in testing right now, c)
podman itself is a much more popular package than nomad-driver-podman
(or nomad for that matter), are all very strong points in favor of
option (2), but are all *on top* of the original point -- again, IMHO.

Best,
Faidon

1: https://github.com/hashicorp/nomad-driver-podman/releases/tag/v0.2.0
2: https://github.com/hashicorp/nomad-driver-podman/issues/69


Reply to: