| 1 |
Joby Walker said: |
| 2 |
> Couple things, |
| 3 |
> |
| 4 |
> 1) I forgot to mention that currently baselayout cannot be packaged, it |
| 5 |
> must be "built" on each server. Annoying but not a huge deal, since the |
| 6 |
> ebuild is mostly file swaping. |
| 7 |
|
| 8 |
Good to know. Not too big a deal, I guess. What is the reason for this, if |
| 9 |
I may ask? I see every server of any given particular architecture all |
| 10 |
having the same "base layout", but I'm sure I'm looking at it at too high |
| 11 |
a level. |
| 12 |
|
| 13 |
Assuming that we have a goal for our deployment solution that we must |
| 14 |
preserve the ability to control package versioning, update source, and |
| 15 |
centralized build process/source, what is the best way to proceed in a |
| 16 |
hybrid sort of way? For example, we could perform the initial server build |
| 17 |
from a stage2 tarball, and for the emerge system afterwards, emerge (up to |
| 18 |
and including) sys-apps/baselayout from source. Anything after this would |
| 19 |
be emerged in using -k switch to ensure that uniform builds are obtained |
| 20 |
from the build server. |
| 21 |
|
| 22 |
Or perhaps, for the initial emerge system, everything up to |
| 23 |
sys-apps/baselayout can be emerged from a package, baselayout build |
| 24 |
locally, and everything afterwards picked up again from packages. I guess |
| 25 |
I would really appreciate some best practice advice at this point ;) |
| 26 |
|
| 27 |
> |
| 28 |
> 2) Not running 'emerge sync' from each server is actually harmful to the |
| 29 |
> server. You might be able to use 'emerge regen' to mitigate most of |
| 30 |
> this, but I haven't investigated it that fully (since I switched from |
| 31 |
> NFS /usr/portage to a BINHOST). |
| 32 |
|
| 33 |
Any chance you could expand on this more? It's starting to sound like it |
| 34 |
may be more difficult than expected to maintain a central portage/package |
| 35 |
source for all of our servers. |
| 36 |
|
| 37 |
> |
| 38 |
> 3) The .keep file prevents an emerge clean from deleting an empty |
| 39 |
> directory. |
| 40 |
> |
| 41 |
> Trust me the BINHOST method works much better than /usr/portage shared |
| 42 |
> over NFS. |
| 43 |
|
| 44 |
Thanks for this advice. I feel much better about BINHOST than I did NFS |
| 45 |
anyway. |
| 46 |
|
| 47 |
Which protocols are supported for BINHOST sources? rsync, http, ftp? |
| 48 |
|
| 49 |
Thanks for your time. |
Archives