[Cialug] CentOS 7

[Zachary Kotlarek]

On Jul 9, 2014, at 9:47 AM, Daniel A. Ramaley <daniel.ramaley at DRAKE.EDU> wrote:

> So really, what problem is this trying to solve? I can see both sides of 
> this one. It would be nice to have constant device names, but it is 
> fixing such a rare problem that i'm not sure if it justifies the effort 
> of changing it.


Mostly they’re just moving toward the state-of-the-art-of-2007 wherein visible device names should be stable, user-defined, and in no way tied to the underlying hardware address. It’s very important for things like USB devices and disks where ordering changes all the time it doesn’t make any sense to specifically exclude hardware type A or B from that plan just because the hardware address for it are somewhat stable on some systems where the hardware never changes.

So in answer to “what problem” I’d suggest “the one where the name changes between boots”. It’s more common with things on shared busses than with NICs, but it happens to anything given the right circumstances. There’s no guarantee that cards will always scan in the same order (many boards do, but it’s not required), and certainly if you installed a new card, or even physically re-arranged them, they’ll come up with different interface numbers and you’ll have to re-map them manually. Or imagine a boot disk used on more than one hardware system. Or replacing a motherboard with one that scans in a different order. Or any of 100 other things that could change the device hardware addresses.

Another is the problem is not knowing what “eth0” and “eth1” represent. A router with interfaces named “uplink” and “LAN” or a laptop with interfaces named “wifi” and “ether” are simply easier to administrate in the same way that the hostnames “test.foo.com” and “prod.foo.com” are easier to use than the underlying network addresses.

Another is removing all of the forced naming conventions that relate to underlying hardware type — why does my T1 card have a different device naming convention than my DSL card which is still different from the Ethernet card? Why does my IP and higher-level configuration care that I changed the underlying device type?

Also note that both the names and the matching criteria are user-defined, so if you still want to name things by PCI address instead of MAC address you can do so simply by editing the config file.

What’s the downside to stable, user-defined, path-independent names?

	Zach

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2749 bytes
Desc: not available
URL: <http://cialug.org/pipermail/cialug/attachments/20140709/d856b5da/attachment.bin>