I ran into this issue when I switched to a LEAF version with a 2.6 based kernel and the only way that I found around it was to use a script to assign the interfaces in the way that I wanted them to be. I had tried varying load order of modules etc but never got it as solid as I wanted. If I remember the default assignment changed with releases as well. This is even an issue on a 2 interface router if you want the ports assigned in a certain way. For example only 1 port is gigabit and you want it on the LAN. Worst of all on a multiple interface router if an interface fails the other interfaces are reordered on bootup. I'm pretty sure I reassigned based on MAC address which works OK if all of your addresses are static. I'm travelling right now and can't check.
----- Dave Dillabough On 2012-10-02, at 10:01 PM, "Erich Titl" <erich.t...@think.ch> wrote: > Hi Martin > > at 02.10.2012 14:14, Martin Hejl wrote: >> Hi Erich, >> >>> I felt pretty sure, as I checked the set up more that once. BUt yes, you >>> are right, pulling down the interface shows that indeed the ethernet >>> numbering had nothing to do with the way I am used to. >> Indeed - it was quite a surprise to me at the time too, since one >> expects all kinds of issues when trying a new piece of hardware, but not >> that the network ports are arranged as "eth1 eth3 eth2 eth0" on one >> model (NSA 1040), and "eth2 eth3 eth0 eth1" on the other (NSA 1045)... >> >>> I do not trust in trial and error and feel like there must be a way to >>> forcibly enumerate the interfaces. How did you solve the issue, as this >>> is quite a showstopper. >> We never really solved it - since the assignment to the network ports >> didn't change with different versions of Linux (various versions of >> Leaf, but we also tried RHEL once), we simply labeled the ports with >> little stickers. It didn't look terribly professional, but it worked. > > I can imagine that it does, but what are the effects on, let's say, snmp > statistics on the interfaces and the fact that I want to use the 1G > interfaces on specific connections without rewiring the cabinet :-( > > This was, according to internet search, introduced in kernel 2.6 and > Dell, running into the same wall has published something to address it, > although only for _real_ distros. I am convinced that we need to address > this issue, as IMHO this is even more important in a firewall scenario. > > http://linux.dell.com/files/whitepapers/consistent_network_device_naming_in_linux.pdf > > cheers > > Erich > > ------------------------------------------------------------------------------ > Don't let slow site performance ruin your business. Deploy New Relic APM > Deploy New Relic app performance management and know exactly > what is happening inside your Ruby, Python, PHP, Java, and .NET app > Try New Relic at no cost today and get our sweet Data Nerd shirt too! > http://p.sf.net/sfu/newrelic-dev2dev > ------------------------------------------------------------------------ > leaf-user mailing list: leaf-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/leaf-user > Support Request -- http://leaf-project.org/ ------------------------------------------------------------------------------ Don't let slow site performance ruin your business. Deploy New Relic APM Deploy New Relic app performance management and know exactly what is happening inside your Ruby, Python, PHP, Java, and .NET app Try New Relic at no cost today and get our sweet Data Nerd shirt too! http://p.sf.net/sfu/newrelic-dev2dev ------------------------------------------------------------------------ leaf-user mailing list: leaf-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-user Support Request -- http://leaf-project.org/
