Package: udev
Version: 0.105-4
Severity: wishlist

The persistent device naming in udev does give quite persistently
headaches ...

Since udev, swapping hardware isn't that easy anymore, as the persistent 
device names try to "reserve" device names.

E.g. if some NIC has been confirmed to be broken and is going to be
replaced, udev's /etc/udev/rules.d/z25_persistent-net.rules "reserves"
eth0 for some device with the MAC adress of the broken NIC, so
the "new" (and only one) NIC in that box is being renamed from eth0 to 
"eth1".

I'm also deploying servers by cloning the installed files onto 
different boxes, changing /etc/network/interfaces to the new ip address,
alter /etc/hostname, /etc/mailname and /etc/hosts to the new network
configuration. Of course, the same problems occur (net devices become
renamed).

This "minor" change creates quite a few problems: as the device name "eth0" 
is in use by quite a few applications and configurations 
(/etc/network/interfaces, /etc/default/dhcp3-server, ...), so 
"automatically" renaming the device creates new problems and
so swapping the NIC (or the whole box including its onboard NIC 
due to some other error)

Other applications like e.g. airmon-ng makes udev create new
devices on the fly (with rising interface numbers); 
http://aircrack-ng.org/doku.php?id=airmon-ng is an example
for this.

My suggestion is to enable/disable persistent net device naming via 
some variable in a to-be-created /etc/default/udev, either globally
or for some specific device.


For my own services, I've created a new
/etc/udev/rules.d/z25_persistent-net.rules file:
---cut
# NEVER rename eth*:
KERNEL=="eth*", NAME="$KERNEL"
---cut

This in turn does match a rule in persistent-net-generator.rules 
both not to run /lib/udev/write_net_rules and makes udev believe
that the device has already been renamed.

However, a more userfriendly solution to accomplish basically the same 
thing would be great.


Anders



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to