1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
On 21.04.2012 17:30, Allan Gottlieb wrote: |
5 |
> On Fri, Apr 20 2012, Neil Bothwick wrote: |
6 |
> |
7 |
>> On Fri, 20 Apr 2012 13:22:20 +0100, Neil Bothwick wrote: |
8 |
>> |
9 |
>>>> I'll run the update again today, paying more attention, and |
10 |
>>>> see what happens. |
11 |
>>> |
12 |
>>> What happened is it broke again, with no obvious signs of the |
13 |
>>> cause. conf-update reported only trivial changes to three |
14 |
>>> files. |
15 |
>> |
16 |
>> I've just tried it on my netbook and the same happened, but I |
17 |
>> think I'm closer to the cause. The three files in /etc/pam.d are |
18 |
>> login, passwd and su. After updating, there were ._cfg* versions |
19 |
>> of these files, but no originals, so conf-update just deleted |
20 |
>> them. It turns out these were owned by shadow but now belong to |
21 |
>> pambase. I suspect that pambase installed them as ._cfg versions, |
22 |
>> because the others already existed, then shadow removed the |
23 |
>> originals as they were no longer part of the package. |
24 |
>> |
25 |
>> Whether this is a bug in portage, the ebuilds or conf-update is |
26 |
>> open to debate, but conf-update ought to handle the situation |
27 |
>> better. I'll file a bug later if no one beats me to it. |
28 |
> |
29 |
> First, thanks for the warning. |
30 |
> |
31 |
> There is a bug filed |
32 |
> https://bugs.gentoo.org/show_bug.cgi?id=412721 |
33 |
> |
34 |
> The comments there say that if you run etc-update right after the |
35 |
> emerge all is well (but this isn't sufficient for people who use |
36 |
> screen, detatch, and log out). Someone also mentioned |
37 |
> dispatch-conf working. No one mentioned cfg-update, which I use |
38 |
> (and I believe neil does as well). Could the problem be dependent |
39 |
> on which configuration file updater one uses? |
40 |
> |
41 |
> I have not updated my primary machine. I did update another one |
42 |
> (both machines are ~amd64) including a cfg-update -q, but have not |
43 |
> rebooted it. The secondary can su. This seems to suggest that |
44 |
> cfg-update is sufficient in some cases. |
45 |
> |
46 |
> Am I correct in believing the safe procedure is to add |
47 |
> |
48 |
>> =sys-auth/pambase-20101024-r2 =sys-apps/shadow-4.1.5. |
49 |
> |
50 |
> to /etc/portage/package.mask (or a file in that directory)? |
51 |
> |
52 |
> thanks, allan |
53 |
> |
54 |
|
55 |
Hi, |
56 |
|
57 |
I actually used cfg-update -u on 3 different machines up to now. |
58 |
So cfg-update can't be at the core of that problem. |
59 |
Maybe it's some kind of race-condition or the bug depends on other |
60 |
things too (e.g.: I'm using gnome and gdm also puts some files to |
61 |
/etc/pam.d which maybe mitigate the issue somehow) - pure speculation, |
62 |
though. |
63 |
|
64 |
The syntax for the masking seems to be correct (since shadow-4.1.5-r2 |
65 |
already has hit the tree maybe the problem is solved. Otherwise you |
66 |
would most likely like to mask -r1 and -r2 also). |
67 |
|
68 |
WKR |
69 |
Hinnerk |
70 |
-----BEGIN PGP SIGNATURE----- |
71 |
Version: GnuPG v2.0.19 (GNU/Linux) |
72 |
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ |
73 |
|
74 |
iQEcBAEBAgAGBQJPkwltAAoJEJwwOFaNFkYcuRwH/2FoHs4JwplMRZlSS4dtg388 |
75 |
y82/o4Cu60kgbdC1kHS7d/OXhu5ZHgTH1KhxW3zZZYxSBc6yGlTV4XBnBveEPBQG |
76 |
R7VkBwLMK7kgQewQGBO2GVIVzDlKa2QtZAHTySgqFritZXZeYrpC5FXC+yj3/k3S |
77 |
tpwZ2RcTFjdaCK8fbELRLtFK4DO00+j7Zs+3NvUz33tTSg8RBKh908DX6IRGW557 |
78 |
Ypd1o1X+Ea8RJcPN71Z8k4EGfwOI3nJW/kpttar3NdRfio6Kc7Gb8MYFeMFIGnX2 |
79 |
AVRTu7pfhdlkjR7+BCXm5kpMtcMZmhN1jelOj8lKtrZsC2VRuYbyjsT+1rssO8Q= |
80 |
=CPBN |
81 |
-----END PGP SIGNATURE----- |