1 |
On 20/12/2016 19:50, Heiko Baums wrote: |
2 |
> Am 20.12.2016 um 18:38 schrieb Kai Peter: |
3 |
>> Maybe there are different opinions, but what is cryptic on - as a |
4 |
>> typical one - enp3s0?: |
5 |
>> e - ethernet |
6 |
>> n - network |
7 |
>> p - pci (port) ... |
8 |
>> 3 - ... 3 |
9 |
>> s - slot ... |
10 |
>> 0 - ... 0 |
11 |
> |
12 |
> Think about that yourself again and compare it to - eth0: |
13 |
> |
14 |
> eth - ethernet |
15 |
> 0 - 1st card |
16 |
|
17 |
No. This is incorrect. |
18 |
|
19 |
eth0 is the first card found by software, and not always the one you |
20 |
think it is. |
21 |
|
22 |
Off the top of my head, a few examples of how this can be a problem: |
23 |
|
24 |
- pull and reseat nic cards in a server, junior flunky who does it |
25 |
forgets which card went where and puts them back in the wrong slots |
26 |
- an update to PCI code does discovery ever so sightly differently |
27 |
- plug in a USB wireless or wired nic, can you absolutely *guarantee* |
28 |
that any kernel will *always* find a PCI nic before a USB one? |
29 |
- virtualized servers, where I can (and very much do) add, remove and |
30 |
edit virtual nics all the time. Now which one is the first? |
31 |
|
32 |
|
33 |
So yes, eth0 is far easier for humans to remember, it's also a fragile |
34 |
solution. On a desktop with only one nic all the above problems never |
35 |
happen. But Linux desktops and single-nic servers are not the target |
36 |
market for Linux, and hasn't been for a very long time. The real target |
37 |
is virtual machines, big iron, and embedded devices plus Android. |
38 |
|
39 |
Oddly enough, my 200+ FreeBSD machines all tend to use a naming |
40 |
convention like enp3s0? for nics and disks. Doesn't cause any issues in |
41 |
that world. |
42 |
|
43 |
In all these threads you participate in recently, I'm not seeing any |
44 |
actual real facts from you, or specifics. All I see is you howling |
45 |
(louder than Poetering!) on the other end of an email address about how |
46 |
you don't like changes that are happening. |
47 |
|
48 |
Actually, I think you don't have much clue about the real world and what |
49 |
it takes to run real fleets of machines. |
50 |
|
51 |
So my advice to you is to put up, or shut up. |
52 |
|
53 |
|
54 |
> |
55 |
> I don't think I need to explain which of both is a lot more complicated |
56 |
> and cryptic. |
57 |
> |
58 |
>> Just an example. The real mess with systemd is that it violates the good |
59 |
>> ol' Unix culture. Especially by "capturing" udev. Thanks to Gentoo for |
60 |
>> eudev!!! |
61 |
> |
62 |
> That's also true but not the only problem with systemd. |
63 |
> |
64 |
> Heiko Baums |
65 |
> |
66 |
|
67 |
|
68 |
-- |
69 |
Alan McKinnon |
70 |
alan.mckinnon@×××××.com |