1 |
On Sun, 02 Mar 2014 14:59:49 +0000 |
2 |
Peter Humphrey <peter@××××××××××××.uk> wrote: |
3 |
|
4 |
> On Sunday 02 Mar 2014 12:21:24 yac wrote: |
5 |
> > On Sat, 01 Mar 2014 22:34:45 +0000 |
6 |
> > |
7 |
> > Peter Humphrey <peter@××××××××××××.uk> wrote: |
8 |
> > > Hello list, |
9 |
> > > |
10 |
> > > While wrestling with my owncloud problem I decided to re-install |
11 |
> > > the server box from scratch. It's a little Atom box so I keep a |
12 |
> > > 32-bit chroot on this machine (the one I'm writing this on), |
13 |
> > > nfs-mount the server's packages directory in the chroot and do |
14 |
> > > all the heavy emerge work on the i5 here. The contents |
15 |
> > > of /etc/portage/* are identical in both systems, with minor |
16 |
> > > exceptions detailing --jobs= and the proxies (the server serves |
17 |
> > > the portage tree and is the rsync host, and it can't do that for |
18 |
> > > itself until http- replicator etc. are in place). |
19 |
> > > |
20 |
> > > I got as far as installing a recent stage 3 and adding nfs-utils |
21 |
> > > so that I could export the package directory. Then I ran emerge |
22 |
> > > -eavK world (I'd already rebuilt all the packages in the chroot |
23 |
> > > before zapping the server). You'll never guess what happened. |
24 |
> > > Well, actually, I hope someone can because I'm stumped again. |
25 |
> > > |
26 |
> > > $ OPTIONS="backtrack=100" emerge -pvKe world |
27 |
> > > ... |
28 |
> > |
29 |
> > Please provide the whole output. There should be an explanation why |
30 |
> > there is the conflict. |
31 |
> |
32 |
> OK. I've been over all my groundwork again, making sure of my facts |
33 |
> and running revdep-rebuild, python-updater and perl-cleaner. |
34 |
> |
35 |
> I ran three commands: |
36 |
> 1. emerge -epvK world > emerge.with-K |
37 |
> 2. emerge -epvk world > emerge.with-k |
38 |
> 3. emerge -epv world > emerge.no-Kk |
39 |
> Those three files are attached. (I pasted the STDERR text into the |
40 |
> first one because I couldn't remember the bashism for redirecting |
41 |
> both output streams to a single file.) |
42 |
> The slot conflict occurred in case 1. It did not occur in cases 2 and |
43 |
> 3, but the package list was ordered differently. |
44 |
|
45 |
-K makes portage to use *only* pre-compiled binary packages, so I guess |
46 |
there's the problem. |
47 |
|
48 |
As you say you are not getting conflicting on later commands, the issue |
49 |
is fixed for you, right? |
50 |
|
51 |
> > You can also run the emerge with -at to get some insight into the |
52 |
> > package dependencies. |
53 |
> |
54 |
> Hmm. A bit of a tall order with 330 packages. |
55 |
> |
56 |
> > What is the reason you run with backtrack=100 ? |
57 |
|
58 |
Actually it seems you are not setting anything as portage doesn't read |
59 |
the OPTIONS variable [1]_. |
60 |
|
61 |
.. [1] http://git.overlays.gentoo.org/gitweb/?p=proj%2Fportage.git&a=search&h=HEAD&st=grep&s=OPTIONS |
62 |
|
63 |
> I don't usually set it at all; I only tried it this time to confirm |
64 |
> my expectation that no amount of back-tracking would make any |
65 |
> difference. |
66 |
> |
67 |
|
68 |
|
69 |
--- |
70 |
Jan Matějka | Gentoo Developer |
71 |
https://gentoo.org | Gentoo Linux |
72 |
GPG: A33E F5BC A9F6 DAFD 2021 6FB6 3EBF D45B EEB6 CA8B |