Using yast I have configured two aliases for my network interface. Everything was successful in yast and viewing the config file in /etc/sysconfig/network/ifcfg-eth0 everything looks as it should be.
The problem is these alias interfaces fail when starting up the network either through rebooting or /etc/init.d/network restart. Here is the output:
Setting up network interfaces:
eth0 device: Digital Equipment Corporation DECchip 21140 [FasterNet] (rev 20)
eth0 IP address: 172.31.0.218/23
eth0:1 IP address: 172.31.0.219/23
eth0:2 IP address: 172.31.0.220/23
eth0 failed
I don’t see any messages explaining what the failure is in any log files I’ve checked. The main eth0 adapter is up and running fine but the two aliases are just not there.
Has anyone had any similar problems with aliases? Any suggestions for what I can check?
And how does /etc/sysconfig/network/ifcfg-eth0 look like? How about posting it? And what is your release? 11.1? Do we get a prize for guessing right? Perhaps something to defray my crystal ball running costs?
I encountered an issue where YaST could create the alias line and bring up the alias interface but did not take down the alias interface when the alias was deleted. But editing ifcfg-eth0 by hand works ok.
OK, I figured it out. I also figured out it is always important to post every detail instead of trying to hide information that doesn’t really need to be hidden.
This is a bug in yast in my opinion.
When I configured the aliases in yast I gave them longer-ish names to match the DNS host names I had set up for them. This is apparently the cause of the failure. There appears to be a length limitation on what the labels can be set to. I edited /etc/sysconfig/network/ifcfg-eth0 and changed the labels for the two aliases to actually be “0” and “1”. After that I restarted the network and everything came up fine.
I think yast should limit you or at least warn you if you enter an alias label that is too long. It would also be nice if the output from starting the network would tell you more than just “failed” or a message in /var/log/messages, or something that at least gives someone a chance to figure out what the problem is.
Oh well, if anyone else runs in to this hopefully this post helps.
Apparently the need for alias names in ifcfg-ethX is going away anyway, this is what I gather from the response to my bug report at Bugzilla, since the ip utility can assign additional IP addresses to an interface without the need for an alias name. I think it’s just waiting for YaST to catch up in the next release. Or so I gathered, I didn’t fully understand the developer comments.
