Re: Ethernet interface numbering in etch

> "Russell" == Russell Coker <[EMAIL PROTECTED]> writes: Russell> For the rare case of a Xen instance with multiple Russell> Ethernet devices it would be easy to modify the config Russell> file in question - which is actually an easier task than Russell> determining how to corr

Re: Ethernet interface numbering in etch

On Sun, Apr 01, 2007 at 10:24:26AM +1100, Russell Coker wrote: > > I disagree. Not only because the bug is not RC, but because you could say > > the same for users running other virtualization technologies (UML? Vmware?) > > with similar behaviours. > > Do they behave in the same way? Well, not t

Re: Ethernet interface numbering in etch

On Saturday 31 March 2007 10:17, Javier Fernández-Sanguino Peña <[EMAIL PROTECTED]> wrote: > If this is so important, why, after so many many months of freeze, no Xen > users found this and submited a bug report? (Now its #413601, filed 24 days > ago) Probably because it's dependent on the order

Re: Ethernet interface numbering in etch

On Mar 31, Javier Fernández-Sanguino Peña <[EMAIL PROTECTED]> wrote: > I'm in favor of adding a note in the Release Notes but I think we should not > delay the release (*again*) by modifying such a critical element as udev > right now. I think the fix would be harmless, but it's too late anyway.

Re: Ethernet interface numbering in etch

On Fri, Mar 30, 2007 at 05:19:25AM -0400, Nathanael Nerode wrote: > Anyone have any thoughts on where and how this should be documented? > Which documents do we need to make patches for and how detailed should > the description in each document be? I think the appropiate location is the Release No

Re: Ethernet interface numbering in etch

On Thu, Mar 29, 2007 at 11:34:53AM +1100, Russell Coker wrote: > The current functionality in regard to Xen is seriously broken. A reasonable > person will expect that when a Xen virtual machine is configured with a > single Ethernet interface then it can be restarted at any time and get the >

Re: Ethernet interface numbering in etch

I think everyone missed the important part of my message. so I'm soliciting comments on it again. >But perhaps the best "solution" is to document prominently that if you >replace your network hardware, you should delete the line associated >with the removed hardware from >/etc/udev/rules.d/

Re: Ethernet interface numbering in etch

On Wednesday 28 March 2007 22:23, "Marco d'Itri" <[EMAIL PROTECTED]> wrote: > > Maybe we could have a special-case for MAC addresses in that range and > > have udev not rename them unless satisfying some other requirement forced > > it. > > It's even easier, just add this rule: > > DRIVERS=="vif",

Re: Ethernet interface numbering in etch

On Wed, Mar 28, 2007 at 06:50:37PM -0400, Joey Hess wrote: > Javier Fernández-Sanguino Peńa wrote: > > I would be interested in documenting how you can do a once-only boot with > > the > > kernel (or linking to available documentation) for the Release Notes. > > grub-rebot N (which uses savedefau

Re: Ethernet interface numbering in etch

Javier Fernández-Sanguino Peña wrote: > I would be interested in documenting how you can do a once-only boot with the > kernel (or linking to available documentation) for the Release Notes. grub-rebot N (which uses savedefault --once) -- see shy jo signature.asc Description: Digital signature

Re: Ethernet interface numbering in etch

On Wed, Mar 28, 2007 at 11:51:15PM +0200, Gabor Gombas wrote: > If you do not have physical access to the machine then serial access is > a must (or some alternative, like IPMI-emulated serial console if you > have the hardware). For example if you have to update the kernel and > something goes wro

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 08:46:26AM -0400, Matthias Julius wrote: > In case of a rented dedicated server I would never enter the machine > room and the person changing hardware would have no login on it. > > For other machines the only possible access is by SSH which is > difficult if the network

Re: Ethernet interface numbering in etch

On Mar 28, Russell Coker <[EMAIL PROTECTED]> wrote: > On Tuesday 27 March 2007 07:11, Luigi Gangitano <[EMAIL PROTECTED]> wrote: > > 2. XEN domUs without fixed mac address setting (configurable in   > > virtual domain configuration file), can have a different MAC address   > > each time are booted

Re: Ethernet interface numbering in etch

On Tuesday 27 March 2007 07:11, Luigi Gangitano <[EMAIL PROTECTED]> wrote: > 2. XEN domUs without fixed mac address setting (configurable in   > virtual domain configuration file), can have a different MAC address   > each time are booted and udev will keep adding a new interface each   > time. Ma

Re: Ethernet interface numbering in etch

I demand that Matthias Julius may or may not have written... > Ben Hutchings <[EMAIL PROTECTED]> writes: >> And almost every laptop comes with wired and wireless interfaces. > I almost wrote that myself, but they have different names and dont > compete for numbers. It depends on which driver is

Re: Ethernet interface numbering in etch

On Tuesday 27 March 2007 11:55, Matthias Julius wrote: > I almost wrote that myself, but they have different names and dont > compete for numbers. That's not necessarily true. My IPW2200 has eth2 on my laptop. wt -- Warren Turkal, Research Associate III/Systems Administrator Colorado State Unive

Re: Ethernet interface numbering in etch

Ben Hutchings <[EMAIL PROTECTED]> writes: > And almost every laptop comes with wired and wireless interfaces. I almost wrote that myself, but they have different names and dont compete for numbers. Matthias -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble

Re: Ethernet interface numbering in etch

On Monday 26 March 2007 16:00, Bernd Zeimetz wrote: > you could mount them by UUID instead of the device name. Don't forget having to use a static fsid parameter in /etc/exports for the affected mounts. Otherwise, the device major/minor numbers for the mounted devices still affect what machines

Re: Ethernet interface numbering in etch

Gabor Gombas <[EMAIL PROTECTED]> writes: > On Mon, Mar 26, 2007 at 11:11:04AM -0500, John Goerzen wrote: > >>configuration, it will no longer bring up the network on boot because >>the device name changed. If the box is using NFS, NIS, or LDAP, >>people may even have trouble logging i

Re: Ethernet interface numbering in etch

On Tue, 2007-03-27 at 08:32 -0400, Matthias Julius wrote: > Mike Hommey <[EMAIL PROTECTED]> writes: > > > On Tue, Mar 27, 2007 at 12:06:09AM -0400, Nathanael Nerode <[EMAIL > > PROTECTED]> wrote: > >> Specifically because: > >> * Most machines have only one interface (If Debian is running on more

Re: Ethernet interface numbering in etch

Mike Hommey <[EMAIL PROTECTED]> writes: > On Tue, Mar 27, 2007 at 12:06:09AM -0400, Nathanael Nerode <[EMAIL > PROTECTED]> wrote: >> Specifically because: >> * Most machines have only one interface (If Debian is running on more >> routers >> than workstations, obviously this would be wrong, but

Re: Ethernet interface numbering in etch

On Mar 27, Nathanael Nerode <[EMAIL PROTECTED]> wrote: > * Most machines have only one interface (If Debian is running on more routers > than workstations, obviously this would be wrong, but I doubt that's the > case.) Every laptop with wifi has more than one interface, and so does every system

Re: Ethernet interface numbering in etch

On Mar 26, Julien Cristau <[EMAIL PROTECTED]> wrote: > I think the problem is that you can't know in advance whether the device > still exists or not, and whether it will be plugged in later (because > everything runs asynchronously). Or if it does not exist because it has been phisically removed

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 12:06:09AM -0400, Nathanael Nerode wrote: > I would guess the second, or more specifically I would guess that in any > given > month, > - (the total number of instances of admins swapping out network hardware and > needing to reuse the same interface names) > is gre

Re: Ethernet interface numbering in etch

On 3/27/07, Steve Langasek <[EMAIL PROTECTED]> wrote: Which means that your network interface wouldn't come up automatically when you changed hardware either, so that's not exactly better than the udev solution in that respect. :) Seems to me udev is only tangentially related to this discussio

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 12:06:09AM -0400, Nathanael Nerode <[EMAIL PROTECTED]> wrote: > After writing a very long message, I realize that there was a much simpler > solution, so if you want to cut to the chase, skip to the end! > > Steve Langasek wrote: > >Which do you think is the common case -

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 06:39:55AM +0200, NN_il_Confusionario wrote: > On Tue, Mar 27, 2007 at 12:01:13AM +0200, Gabor Gombas wrote: > > No, ever since distros started using modular kernels, hotplug _is_ the > > norm. > debian (and other distros) used modular kernels (2.4.18 in woody) > without ho

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 12:06:09AM -0400, Nathanael Nerode wrote: > After writing a very long message, I realize that there was a much simpler > solution, so if you want to cut to the chase, skip to the end! > Steve Langasek wrote: > >Which do you think is the common case -- a system with more th

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 12:01:13AM +0200, Gabor Gombas wrote: > No, ever since distros started using modular kernels, hotplug _is_ the > norm. debian (and other distros) used modular kernels (2.4.18 in woody) without hotplug or the like. > You can get rid of it only by building your own kernel im

Re: Ethernet interface numbering in etch

After writing a very long message, I realize that there was a much simpler solution, so if you want to cut to the chase, skip to the end! Steve Langasek wrote: >Which do you think is the common case -- a system with more than one network >interface where it's necessary to preserve interface order

Re: Ethernet interface numbering in etch

I demand that Luigi Gangitano may or may not have written... [snip] > 1. Some ethernet cards like Sun QuadFE share the same MAC address (even if > global OpenFirmware option is set to different MAC-address) and PCI id and > udev blocks while renaming them, leaving with an unusable systema each tim

Re: Ethernet interface numbering in etch

> Btw. do Debian initrds already support specifying the root fs with > LABEL= like Fedora/RedHat? > Didn't try it, but according to [1] they do. Cheers, Bernd [1] http://wiki.debian.org/InitrdReplacementOptions -- Bernd Zeimetz <[EMAIL PROTECTED]>

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 11:11:04AM -0500, John Goerzen wrote: > * dmesg output still mentions hardware using eth0, even if you can't >talk to it at eth0 but must instead use eth5. dmesg doesn't >mention this fact, making it difficult to track down problems. It's nothing new, this is the

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 01:41:26PM -0500, John Goerzen wrote: > On Mon, Mar 26, 2007 at 07:54:45PM +0200, Hendrik Sattler wrote: > > > The trouble is that nowadays, the kernel does not assign predictable > > > interface names, and an increasing number of systems has got more than > > > one Ethernet

Re: Ethernet interface numbering in etch

On Tue, Mar 27, 2007 at 12:00:05AM +0200, Bernd Zeimetz wrote: > you could mount them by UUID instead of the device name. Btw. do Debian initrds already support specifying the root fs with LABEL= like Fedora/RedHat? Gabor -- - M

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 01:40:14PM -0500, John Goerzen wrote: > Still the exception, not the norm. No, ever since distros started using modular kernels, hotplug _is_ the norm. You can get rid of it only by building your own kernel image with all hardware drivers statically built in and module sup

Re: Ethernet interface numbering in etch

Heya, > As a corollary to this, I have machines where the disks swap device files on > each boot. It's pretty annoying when my nfs volumes switch which device name > is used to mount them. > you could mount them by UUID instead of the device name. Bernd -- Bernd Zeimetz <[EMAIL PROTECTED]>

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 11:11:32PM +0200, Luigi Gangitano wrote: > BTW, there's no easy way to recover from a badly renamed ethernet interface. > Once you have something like 'eth5_rename' how are you supposed to recover? As you would do for a "normal" interface rename: ip link set eth5_

Re: Ethernet interface numbering in etch

Florian Weimer <[EMAIL PROTECTED]> writes: > * Russ Allbery: >> There's actually some stuff in udev or some related package to deal >> with this, but I can't ever seem to find it when I need it. I think >> this is actually a documentation bug more than a functionality bug; we >> just need a bette

Re: Ethernet interface numbering in etch

Hi John! On Mon, 26 Mar 2007, John Goerzen wrote: > On Mon, Mar 26, 2007 at 08:28:02PM +0200, Bastian Blank wrote: > > On Mon, Mar 26, 2007 at 08:15:30PM +0200, Florian Weimer wrote: > > > Sure, but hotpluggable PCI(e) interfaces are the exception, not the > > > norm. It seems wrong to optimize

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 11:11:32PM +0200, Luigi Gangitano wrote: > 1. Some ethernet cards like Sun QuadFE share the same MAC address > (even if global OpenFirmware option is set to different MAC-address) > and PCI id and udev blocks while renaming them, leaving with an > unusable systema eac

Re: Ethernet interface numbering in etch

Il giorno 26/mar/07, alle ore 21:29, Mark Brown ha scritto: The use cases where users are likely to notice are relatively limited - you need to either be trying to do some sort of system imaging or doing hardware replacement where you need to do a like for like swap. The latter case tends to

Re: Ethernet interface numbering in etch

* Bastian Blank: > On Mon, Mar 26, 2007 at 08:15:30PM +0200, Florian Weimer wrote: >> Sure, but hotpluggable PCI(e) interfaces are the exception, not the >> norm. It seems wrong to optimize for this case. > > udev sees network devices, not pci devices. and hotpluggable network > devices are commo

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 01:41:26PM -0500, John Goerzen wrote: > On Mon, Mar 26, 2007 at 07:54:45PM +0200, Hendrik Sattler wrote: > > You only need to delete /etc/udev/rules.d/z25_persistent*.rules before udev > > runs. That should be doable and could be a configuration option. The > > default >

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 07:54:45PM +0200, Hendrik Sattler wrote: > > The trouble is that nowadays, the kernel does not assign predictable > > interface names, and an increasing number of systems has got more than > > one Ethernet interface. The downside is that typical Debian > > installations are

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 08:28:02PM +0200, Bastian Blank wrote: > On Mon, Mar 26, 2007 at 08:15:30PM +0200, Florian Weimer wrote: > > Sure, but hotpluggable PCI(e) interfaces are the exception, not the > > norm. It seems wrong to optimize for this case. > > udev sees network devices, not pci devic

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 08:15:30PM +0200, Florian Weimer wrote: > Sure, but hotpluggable PCI(e) interfaces are the exception, not the > norm. It seems wrong to optimize for this case. udev sees network devices, not pci devices. and hotpluggable network devices are common. Bastian -- A woman sh

Re: Ethernet interface numbering in etch

* Julien Cristau: > I think the problem is that you can't know in advance whether the device > still exists or not, and whether it will be plugged in later (because > everything runs asynchronously). Sure, but hotpluggable PCI(e) interfaces are the exception, not the norm. It seems wrong to opti

Re: Ethernet interface numbering in etch

Am Montag 26 März 2007 19:26 schrieb Florian Weimer: > * Russ Allbery: > > There's actually some stuff in udev or some related package to deal with > > this, but I can't ever seem to find it when I need it. I think this is > > actually a documentation bug more than a functionality bug; we just nee

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 12:42:54 -0500, John Goerzen wrote: > On Mon, Mar 26, 2007 at 07:35:39PM +0200, Marco d'Itri wrote: > > On Mar 26, John Goerzen <[EMAIL PROTECTED]> wrote: > > > > > I think that the right thing to do is to assign the persistent names to > > > network devices that still exi

Re: Ethernet interface numbering in etch

On Mon, Mar 26, 2007 at 07:35:39PM +0200, Marco d'Itri wrote: > On Mar 26, John Goerzen <[EMAIL PROTECTED]> wrote: > > > I think that the right thing to do is to assign the persistent names to > > network devices that still exist in the system, but to do nothing with > > any other network devices.

Re: Ethernet interface numbering in etch

On Mar 26, John Goerzen <[EMAIL PROTECTED]> wrote: > I think that the right thing to do is to assign the persistent names to > network devices that still exist in the system, but to do nothing with > any other network devices. That will allow systems to still boot and > come up properly in the fa

Re: Ethernet interface numbering in etch

* Russ Allbery: > There's actually some stuff in udev or some related package to deal with > this, but I can't ever seem to find it when I need it. I think this is > actually a documentation bug more than a functionality bug; we just need a > better guide on how to do it. You can, somehow, assig

Re: Ethernet interface numbering in etch

John Goerzen <[EMAIL PROTECTED]> writes: > I have spent the past few days trying to figure out why some of our > machines seem to have ethernet interface numbers that jump around -- > eth0 one day, then eth4 or eth5 another. > The culprit comes down to udev. I've filed a bug #416284 against it f

Re: Ethernet interface numbering in etch

On Monday 26 March 2007 10:11, John Goerzen wrote: > I have spent the past few days trying to figure out why some of our > machines seem to have ethernet interface numbers that jump around -- > eth0 one day, then eth4 or eth5 another. As a corollary to this, I have machines where the disks swap de

Ethernet interface numbering in etch

Hi, I have spent the past few days trying to figure out why some of our machines seem to have ethernet interface numbers that jump around -- eth0 one day, then eth4 or eth5 another. The culprit comes down to udev. I've filed a bug #416284 against it for this. Basically, udev is trying to assign