it is just plain not safe you guys update your scripts ofen enough.. (even in stable) that this behavior occurs on production systems.. and Whoops! a new net connection is started! The error message should read: net.eth1 config not found, skipping, to start a temp dhcp session run appropriate-temp-net-script-goes-here eth1 or similar Thanks
personally i see this as being a dupe of Bug 70131 leaving the current default makes sense to me
I don't see that as a dupe. There is nothing "not safe" about using dhcp - it will either work or fail. Nor is it a security risk imo. If you don't want the interface to have any configuration you can do in /etc/conf.d/net config_eth0=( "null") Or stop init scripts from being coldplugged or hotplugged, which is where I suspect the root of the problem is. in /etc/conf.d/rc RC_COLDPLUG="no" RC_HOTPLUG="no"
I didn't even add net.eth1 to my list of items to start... so maybe that's where the bug is... the system is trying to start it! net.eth0 default net.eth1 I guess I need to summarize the bug... I guess dhcp isn't a bad default... for a service that has been asked to start.. but I did not ask this service to start! I did not do the rc-update add net.eth1 default Most linux systems (at least in the past) required the network to be turned on.. I did not turn this network on!
sounds like hotplug/coldplug or maybe udev crap again ... are you perhaps using any of these three things ?
(In reply to comment #3) > I didn't even add net.eth1 to my list of items to start... so maybe that's > where the bug is... the system is trying to start it! I told you already how to fix that. (In reply to comment #2) > Or stop init scripts from being coldplugged or hotplugged, which is where I > suspect the root of the problem is. > > in /etc/conf.d/rc > RC_COLDPLUG="no" > RC_HOTPLUG="no" >
it's a bug.. I use udev hotplug and coldplug I'm adding those to the summary. It's not ok to start that script... and wierd behavior like this is just not acceptable. how do I tell which one it is.. without rebooting?
I don't use coldplug.. it's not even installed.
It's udev doing the coldplugging for you - which you probably do have installed. *** This bug has been marked as a duplicate of 119989 ***
even if udev was trying to start this service.. it shouldn't start... if you guys want to have things start by default have the /etc/init.d/net.eth0 and net.eth1 (which I had to create.. so I'm noticing a bug that others wouldn't see untill it bites them) configured to start the device by default... If the config is not set, and the rc scripts to ask for it to start... , the service should not start... it should be prevented from starting even. You can make a default config, and still keep all the newbies happy...
(In reply to comment #9) > even if udev was trying to start this service.. it shouldn't start... This is why we have RC_COLDPLUG="yes|no" to control wether we allow this or not. By default we *do* allow this. It is called coldplugging.