1 |
Hi all, |
2 |
|
3 |
Funny you talk about this, cause i'm gonna be developing a similar |
4 |
system starting next Semptember as part of the final project for my |
5 |
graduation in Informatics Engineering (kinda Computer Science but |
6 |
that's the official name). My project will be about deploying Gentoo |
7 |
in a not so large environment (few dozen machines) and with similar |
8 |
objectives. So i'd love to exchange ideas to anyone who had this |
9 |
previous experience or is planning such a thing. |
10 |
|
11 |
As to your questions, having a custom portage will be easy, just |
12 |
insert what you need in a file and insert that file at |
13 |
RSYNC_EXCLUDEFROM in make.conf. Be careful with depenencies. |
14 |
|
15 |
Not having gcc will be a problem, as it's part of the base profile, |
16 |
from which all profiles inherit. I don't know if modifying it for |
17 |
your local systems will be a valid alternative. |
18 |
|
19 |
Keeping binary packages is easy as long as you compile them with a |
20 |
more wider set of flags that will be adequate to all machines, like |
21 |
compiling them to i686. Problems will arrive if you have different |
22 |
arch's there. Portage will happily work with binary packages, even |
23 |
fetching them from a primary server using http or ftp. |
24 |
|
25 |
My approach will be slightly different, i'm planing on having a |
26 |
central portage database (a relational database, mysql or postgres) |
27 |
and make a modified portage that will make all queries on this |
28 |
database, to avoid rsync. Will be checking the (dead?) portagesql |
29 |
project when I start. |
30 |
|
31 |
Hope it helps. |
32 |
|
33 |
Ricardo Loureiro |
34 |
|
35 |
On Thu, 30 Jun 2005 19:18:02 -0700 |
36 |
"Joe Rizzo" <jrizzo@×××××××××.com> wrote: |
37 |
|
38 |
> I hope I am posting this to the correct mailing lists. The |
39 |
> gentoo-servers and gentoo-admin mailing lists seem appropriate. |
40 |
> |
41 |
> I am looking at deploying a large scale installation of systems |
42 |
> running Gentoo. (Large scale being 200-300 systems) Currently, I |
43 |
> am mainly concerned with how to handle portage and packages. |
44 |
> |
45 |
> My current thinking is to have a limited portage tree that only |
46 |
> contains ebuilds needed by the systems. Included in this portage |
47 |
> tree will be ebuilds for custom software packages. The systems will |
48 |
> sync off of this custom maintained portage tree. I would like |
49 |
> binary packages to be available from a central repository and not |
50 |
> have gcc on the systems. I would like to avoid mounting a network |
51 |
> file system. Am I on the best path? |
52 |
> |
53 |
> Functionally, systems will be imaged the minimal base image via |
54 |
> systemimager. After that, packages will be deployed via emerge. |
55 |
> The packages will be precompiled and available. On going, packages |
56 |
> will be updated and deployed via emerge. |
57 |
> |
58 |
> Please provide experience or ideas on: |
59 |
> 1) Creating and maintaining a minimal gentoo image? |
60 |
> 2) Managing portage and packages for a large scale gentoo |
61 |
> environment? |
62 |
> |
63 |
> Thanks, |
64 |
> Joe Rizzo |
65 |
> |
66 |
> -- |
67 |
> gentoo-admin@g.o mailing list |
68 |
> |
69 |
|
70 |
-- |
71 |
http://pgp.dei.uc.pt:11371/pks/lookup?op=get&search=0x6B7C0EC0 |