Hello,
this bug occurs on this sarge system, too. I'm using ifplugd in combination
with ifrename. A ps aux | grep ifplugd reveals not only a running ifplugd on
the renamed ether0 interface, but also for the no longer existant eth0
device.
stracing this process reveals what ifplugd does that is so time-consuming:
strace -f -tt -o /tmp/ifplugd /usr/sbin/ifplugd -i eth0 -q -f -u0 -d3 -w -I:
[...]
16706 22:49:19.754452 select(1024, [5], NULL, NULL, {1, 0}) = 0 (Timeout)
16706 22:49:20.754192 ioctl(7, SIOCGIFFLAGS, 0xbffafae0) = -1 ENODEV (No such
device)
16706 22:49:20.786811 ioctl(7, SIOCETHTOOL, 0xbffafb40) = -1 ENODEV (No such
device)
16706 22:49:20.818610 ioctl(7, SIOCGIFFLAGS, 0xbffafaf0) = -1 ENODEV (No such
device)
16706 22:49:20.850862 ioctl(7, SIOCGMIIPHY, 0xbffafb40) = -1 ENODEV (No such
device)
16706 22:49:20.882136 ioctl(7, SIOCGIFFLAGS, 0xbffafae0) = -1 ENODEV (No such
device)
16706 22:49:20.914679 ioctl(7, SIOCGIWAP, 0xbffafb30) = -1 ENODEV (No such
device)
16706 22:49:20.946584 ioctl(7, SIOCGIFFLAGS, 0xbffafaf0) = -1 ENODEV (No such
device)
16706 22:49:20.979133 ioctl(7, SIOCGIFFLAGS, 0xbffafb40) = -1 ENODEV (No such
device)
16706 22:49:21.009833 select(1024, [5], NULL, NULL, {1, 0}) = 0 (Timeout)
16706 22:49:22.009300 ioctl(7, SIOCGIFFLAGS, 0xbffafae0) = -1 ENODEV (No such
device)
16706 22:49:22.042169 ioctl(7, SIOCETHTOOL, 0xbffafb40) = -1 ENODEV (No such
device)
16706 22:49:22.073503 ioctl(7, SIOCGIFFLAGS, 0xbffafaf0) = -1 ENODEV (No such
device)
16706 22:49:22.105409 ioctl(7, SIOCGMIIPHY, 0xbffafb40) = -1 ENODEV (No such
device)
16706 22:49:22.137864 ioctl(7, SIOCGIFFLAGS, 0xbffafae0) = -1 ENODEV (No such
device)
16706 22:49:22.170268 ioctl(7, SIOCGIWAP, 0xbffafb30) = -1 ENODEV (No such
device)
16706 22:49:22.201533 ioctl(7, SIOCGIFFLAGS, 0xbffafaf0) = -1 ENODEV (No such
device)
16706 22:49:22.233258 ioctl(7, SIOCGIFFLAGS, 0xbffafb40) = -1 ENODEV (No such
device)
16706 22:49:22.266295 select(1024, [5], NULL, NULL, {1, 0}) = 0 (Timeout)
[...]
So, there are probably two bugs. The first: ifplugd should not be started for
the eth0 interface, which is there only for a short time before ifrename
kicks in. The second: Once running, ifplugd should detect that the device
certainly is not there and at least after the first timeout it should stop
trying.
If you need the full log or more info, do not hesitate to contact me.
-Malte Cornils
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]