Bug 112088 – Network dependant services are not re-started after restarting a network interface

• Product: Gentoo Linux
• Status: VERIFIED FIXED

[Description Henrik Brix Andersen 2005-11-10 10:27:43 UTC]

Starting with sys-apps/baselayout-1.12.0_pre10-r1, the network dependant
services which are stopped before bringing down the interface when issuing
`/etc/init.d/net.eth0 restart` are not restarted when the network is brought
back up in the background.

[Comment 1 SpanKY 2005-11-10 10:36:56 UTC]

this has been filed already, just not sure of the bug number

[Comment 2 Henrik Brix Andersen 2005-11-10 10:41:44 UTC]

I couldn't find any relevant bug - and Uberlord asked me to open this.

[Comment 3 SpanKY 2005-11-10 10:50:38 UTC]

looks like Bug 80785 to me

[Comment 4 Henrik Brix Andersen 2005-11-10 11:04:57 UTC]

(In reply to comment #3)
> looks like Bug 80785 to me

No, that deals with bringing down the interface and then later bringing it back up.

I'm talking about restarting the interface using `/etc/init.d/net.eth0 restart`.
This works in <1.12.0_pre10-r1, but the immediate backgrounding in _pre10-r1
breaks this behavior.

[Comment 5 Roy Marples (RETIRED) 2005-11-10 14:22:42 UTC]

I'm in two minds about this.

If you restart a network service that brings itself up in the background (ie it
returns straight away allowing for a faster boot) then should it restart the
depending services when it comes up? One solution would be to create a move the
snapshot.$$ folder to snapshot.$service if $service is inactive so when $service
is "up" then it can restart things ok.

On the other hand, there is no guarantee that the service will come up.

[Comment 6 Henrik Brix Andersen 2005-11-10 14:46:14 UTC]

Well, this behavior changed in pre10-r1. Why not just restore the behavior of pre10?

[Comment 7 Roy Marples (RETIRED) 2005-11-10 14:55:17 UTC]

Actually we just changed the default timeouts to -1 for ifplugd, netplugd and
wpa_supplicant

[Comment 8 Roy Marples (RETIRED) 2005-11-15 03:25:29 UTC]

The issue here is the default timeout.

-1 returns instantly, so the interface is "inactive" and not "started". If
enough people are unhappy with this behaviour then we will change the default
back to 60 seconds.

[Comment 9 Roy Marples (RETIRED) 2005-11-17 00:42:45 UTC]

*** Bug 112762 has been marked as a duplicate of this bug. ***

[Comment 10 Roy Marples (RETIRED) 2005-11-23 02:53:58 UTC]

*** Bug 113312 has been marked as a duplicate of this bug. ***

[Comment 11 Roy Marples (RETIRED) 2006-01-10 08:17:03 UTC]

Fixed in SVN -will be in baselayout-1.12.0_pre14

[Comment 12 Roy Marples (RETIRED) 2006-01-13 05:04:31 UTC]

pre14 is out

[Comment 13 Henrik Brix Andersen 2006-01-13 10:17:29 UTC]

Verified with baselayout-1.12.0_pre14. Thanks :)