On 07/08/2010 11:37 AM, Agblad Tore wrote:
> Hi, we have a strange problem with adding an extra network adapater in a 
> SLES11 SP1
> 
> Starting with one works fine.
> Adding one more always ends up with the first ip-address not posible to login 
> via ssh
>   but the added nr 2 ip-address works, so we can login ( thank's for that )

That seems independent from the renaming below which works fine.
Could it be, that for your second interface also gets a default route
configured?
If so, typically, the last one wins and renders the other still existing
default routes in the routing table disabled (unless using some kind of
round robin load balancing).
Unfortunately, I don't know from the top of my head what needs to be
changed in the ifcfg files.

> We have the following def in z/VM  ( 5.4 )
> 
> NICDEF 0700 TYPE QDIO DEV 3 LAN SYSTEM VSW1
> NICDEF 0710 TYPE QDIO DEV 3 LAN SYSTEM VSW1
> 
> below is from boot.msg, there is a strange rename, why ??

> even more strange, after that rename in boot.msg, further down in same 
> boot.msg
> it displays eth0 as ip ending with 122 and eth1 as ip ending with 115 which is
> the order we have configured them into, so what rename ???

Not sure if the file name is exactly correct for SLES 11 but there's
something like a /etc/udev/rules.d/70-persistent-network.rules (with
SLES 10 it is 30-net_persistent_names.rules).
For each network device that appears and is not already in that rules
file, some udev script inserts a new entry with the next free available
device name.
For each network device that is already in that rules file, it causes
Linux network devices/interfaces that come up (during boot or
dynamically) to be renamed to a stable persistent name.
On s390x, the key for matching a device is the device bus ID as opposed
to the MAC address as on other platforms.
If you would like to change the persistent names, you may edit that
rules file and change the ethX to your liking. Each persistent device
name must be unique in the Linux instance.

In your case, the "second" interface 0710 happens to be set online first
and qeth allocates eth0, then udev renames it to eth1 since it knows
that your "first" interface 0700 should be eth0 and will need that name
when that one is brought online afterwards.
Basically, the persistent names work around the indeterminisms in udev
when loading device drivers or establishing ccwgroups and setting them
online.

The ifcfg files use device names as key to apply the configuration to
the correct interface.

> <6>qeth.736dae: 0.0.0710: Device is a Guest LAN QDIO card (level: V543)
> <4>with link type GuestLAN QDIO (portname: ZLINUX1)

> <6>eth0 renamed to eth1 by udevd [676]             
> <------------------------------ STRANGE RENAME HERE
> <6>udev: renamed network interface eth0 to eth1

> <6>qeth.736dae: 0.0.0700: Device is a Guest LAN QDIO card (level: V543)
> <4>with link type GuestLAN QDIO (portname: ZLINUX)

> If we remove ( or delete in Yast 710 'not configured' )
> the first 700 with ip ending with 122 starts working again.

Steffen

Linux on System z Development

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Martin Jetter
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

Reply via email to