1 |
On Mon, Sep 14, 2020 at 12:30:48PM +0200, Marc Schiffbauer wrote: |
2 |
> * William Hubbs schrieb am 14.09.20 um 00:39 Uhr: |
3 |
> > All, |
4 |
> > |
5 |
> > I would like to get some thoughts on kubernetes packaging. |
6 |
> > |
7 |
> > When I started maintaining it in Gentoo, it was packaged as 7 ebuilds |
8 |
> > (one per executable), and only one of them was marked stable. |
9 |
> > |
10 |
> > Since we normally do not split up monorepos into separate packages, I |
11 |
> > started moving everything over to one kubernetes ebuild. |
12 |
> > Now a bug has |
13 |
> > been opened which has a good case for kubeadm being a package on its |
14 |
> > own, so I have done that [1]. |
15 |
> > |
16 |
> > I need to know the best way to proceed, so I'll throw out a couple |
17 |
> > of questions: |
18 |
> > |
19 |
> > 1) should I bring back the split packages and lastrites |
20 |
> > sys-cluster/kubernetes? |
21 |
> > |
22 |
> > 2) should I just bring back other split packages that need to be split |
23 |
> > as I find them? |
24 |
> > |
25 |
> > What do folks think would be the best way for us to package Kubernetes? |
26 |
> |
27 |
> |
28 |
> Interesting. |
29 |
> |
30 |
> So it seems like at least kubeadm and kube-apiserver need to be in |
31 |
> seperate packages. |
32 |
> |
33 |
> I am not a kubernetes guy, but would SLOTting be an option? Like |
34 |
> postgresql for example where you need both versions, old a new to do |
35 |
> database migration. |
36 |
|
37 |
I'm not really interested in slotting; that becomes complicated really |
38 |
quickly -- renaming all of the binaries to version-specific names |
39 |
writing an eselect module, etc. |
40 |
|
41 |
> If this is not an option I would say this is a case for split package |
42 |
> and perhaps a meta-package bringing all of them together. |
43 |
|
44 |
Thinking about it more, the split packages are going to be the best |
45 |
setup. However, a meta package would only be valuable if all parts of |
46 |
kubernetes are needed on the same machine, and that isn't the case most |
47 |
of the time from what I've been told. |
48 |
|
49 |
If folks think that single case justifies one, let me know. |
50 |
My thought is that the meta package, if I create one, should only have a |
51 |
two level version number and use * dependencies to match all package |
52 |
versions in that range. |
53 |
|
54 |
What do others think? is it worth a meta package for the case of having |
55 |
all kubernetes binaries on one machine? If so, what do you think about |
56 |
my suggestion of the minor versions for the meta package? |
57 |
|
58 |
William |