1 |
2014-07-27 17:50 GMT+03:00 Daniel Frey <djqfrey@×××××.com>: |
2 |
> On 07/27/2014 04:30 AM, Grand Duet wrote: |
3 |
>> |
4 |
>>> and also the output of the "rc-update show" command? |
5 |
>> |
6 |
>> # rc-update show |
7 |
>> alsasound | boot |
8 |
>> bootmisc | boot |
9 |
>> dbus | default |
10 |
>> devfs | sysinit |
11 |
>> dmesg | sysinit |
12 |
>> fsck | boot |
13 |
>> hostname | boot |
14 |
>> hwclock | boot |
15 |
>> keymaps | boot |
16 |
>> killprocs | shutdown |
17 |
>> kmod-static-nodes | sysinit |
18 |
>> local | default |
19 |
>> localmount | boot |
20 |
>> loopback | boot |
21 |
>> metalog | default |
22 |
>> modules | boot |
23 |
>> mount-ro | shutdown |
24 |
>> mtab | boot |
25 |
>> net.eth0 | default |
26 |
>> net.lo | boot |
27 |
>> netmount | default |
28 |
>> privoxy | default |
29 |
>> procfs | boot |
30 |
>> root | boot |
31 |
>> savecache | shutdown |
32 |
>> swap | boot |
33 |
>> swapfiles | boot |
34 |
>> sysctl | boot |
35 |
>> sysfs | sysinit |
36 |
>> termencoding | boot |
37 |
>> tmpfiles.dev | sysinit |
38 |
>> tmpfiles.setup | boot |
39 |
>> udev | sysinit |
40 |
>> udev-mount | sysinit |
41 |
>> urandom | boot |
42 |
>> |
43 |
>> Everywhere above eth0 has been put instead of its udev "predictable" name. |
44 |
>> |
45 |
>> Do you think that I need |
46 |
>> carrier_timeout_eth0=20 |
47 |
>> somewhere in /etc/conf.d/net ? |
48 |
>> |
49 |
> |
50 |
> Have you tried disabling network hotplugging in /etc/rc.conf? |
51 |
> i.e. setting rc_hotplug="!net.*" |
52 |
|
53 |
No, I have rc_hotplug="*" in my /etc/rc.conf file. |
54 |
|
55 |
I will try it, thank you for the tip. |
56 |
|
57 |
> Sounds like the interfaces are being brought up out of order. |
58 |
|
59 |
Yes, I have just written about it the following: |
60 |
|
61 |
I finally got an idea why everything went wrong after |
62 |
the last system update. |
63 |
|
64 |
During my last system update, portage instructed me to add |
65 |
dev-lang/tk-8.5.15 threads |
66 |
line to my /etc/portage/package.use file. |
67 |
|
68 |
Here is the portage message about it: |
69 |
|
70 |
The following USE changes are necessary to proceed: |
71 |
(see "package.use" in the portage(5) man page for more details) |
72 |
# required by dev-lang/ruby-1.9.3_p484[tk] |
73 |
# required by dev-ruby/rake-0.9.6[ruby_targets_ruby19] |
74 |
# required by dev-lang/ruby-2.0.0_p353 |
75 |
# required by dev-ruby/racc-1.4.9[ruby_targets_ruby20] |
76 |
# required by dev-ruby/rdoc-4.0.1-r1[ruby_targets_ruby19] |
77 |
=dev-lang/tk-8.5.15 threads |
78 |
|
79 |
And in my /etc/portage/package.use file this line has already |
80 |
been commented. So, I just uncommented it. But I do remember |
81 |
that it was commented for a reason, though do not remember |
82 |
exactly why. |
83 |
|
84 |
May be uncommenting that line allowed bringing up |
85 |
lo and eth0 interfaces *in parallel*. If it is the case, |
86 |
then I have an easy explanation why the contents |
87 |
of /etc/resolv.conf file after boot is unpredictable. |
88 |
|
89 |
If eth0 starts after lo, then I have the right /etc/resolv.conf |
90 |
file, however if lo starts after eth0, then the DNS IPs in |
91 |
resolv.conf file are overwritten with dummy instruction |
92 |
for lo interface. |
93 |
|
94 |
But, now, after your suggestion, I have looked into |
95 |
my /etc/rc.conf file, and have found there the option |
96 |
rc_parallel="NO" |
97 |
which softens my previous arguments a bit, but not completely: |
98 |
may be lo and eth0 are brought up not in parallel but in different |
99 |
order, anyway. |