Bug 1055345

It looks like a problem with wicked but not sure at all. So logs would be appreciated.

World need advice how to get to the log. 


Meanwhile I have installed another distribution on the same hardware with the same DHCP server. It worked successfully. Seems it has to do with the installer.

Sure, my fault.

You can attach logs as documented here:

https://en.opensuse.org/openSUSE:Report_a_YaST_bug#Attachments_-_y2logs.2C_hwinfo_etc.

And to invoke a terminal or switch to the console:

https://en.opensuse.org/SDB:YaST_tricks#Invoking_terminal_from_YaST

Created attachment 741446 [details]
Installation logs from /var/log

Installation logs as far as can be collected from the early stage of installation.

(In reply to Florian Bilek from comment #0)
> I would like to provide the installation log but I am not able to find it
> back.

(In reply to Florian Bilek from comment #4)
> Installation logs from /var/log

Thanks for the logs!

See https://en.opensuse.org/SDB:Linuxrc

Using debug=4 for linuxrc via kernel parameter like
   linuxrc.debug=4 linuxrc.log=/var/log/YaST2/linuxrc.log
would IMO make sense here and should make more details about dhcp visible.

Probably also "debug.wait=net:60 ifcfg=eth0=dhcp4,DHCLIENT_WAIT_AT_BOOT=30"

> The problem seems to be related to exactly this type of network card.

Yes, this is possible & quite likely, that this is hardware/driver specific.
We install our all (except of corner cases) development/test machines via pxe.

[In fact, it already happened in the past, that a network card using bnx/bnc2
 driver reported carrier, but did not forwarded packets for a quite long while,
 especially with "cold" NICs [kind of long init in background when it goes UP
 first time]. But this was AFAIR solved quite long time ago...]

Does it also happen when you pass static IP setup as kernel parameters
to linuxrc via pxe or only when dhcp is used?

  append initrd=... ifcfg="eth0=10.0.1.1/24,....." install=....

I've just tried a leap-42.3 installation (dhcp in linuxrc of course) on a
machine with BCM5709 NICs, but it just worked without any suspect delays:

morricone:~/:[0]# hwinfo --netcard --short
network:                                                        
  eth0                 Hewlett-Packard Company NC382i Integrated Multi-port PCI Express Gigabit Server Adapter
  eth1                 Hewlett-Packard Company NC382i Integrated Multi-port PCI Express Gigabit Server Adapter
  eth2                 Hewlett-Packard Company NC382i Integrated Multi-port PCI Express Gigabit Server Adapter
  eth3                 Hewlett-Packard Company NC382i Integrated Multi-port PCI Express Gigabit Server Adapter
morricone:~/:[0]# hwinfo --netcard 
35: PCI 300.0: 0200 Ethernet controller                         
  [Created at pci.378]
  Unique ID: rBUF.E2l3+2nwDtC
  Parent ID: CLZK.LYDvw_7sQA1
  SysFS ID: /devices/pci0000:00/0000:00:12.0/0000:03:00.0
  SysFS BusID: 0000:03:00.0
  Hardware Class: network
  Model: "Hewlett-Packard Company NC382i Integrated Multi-port PCI Express Gigabit Server Adapter"
  Vendor: pci 0x14e4 "Broadcom"
  Device: pci 0x1639 "NetXtreme II BCM5709 Gigabit Ethernet"
  SubVendor: pci 0x103c "Hewlett-Packard Company"
  SubDevice: pci 0x7055 "NC382i Integrated Multi-port PCI Express Gigabit Server Adapter"
  Revision: 0x20
  Driver: "bnx2"
  Driver Modules: "bnx2"
  Device File: eth0
  Memory Range: 0xf6000000-0xf7ffffff (rw,non-prefetchable)
  IRQ: 19 (no events)
  HW Address: 00:23:7d:60:71:ee
  Permanent HW Address: 00:23:7d:60:71:ee
  Link detected: no
  Module Alias: "pci:v000014E4d00001639sv0000103Csd00007055bc02sc00i00"
  Driver Info #0:
    Driver Status: bnx2 is active
    Driver Activation Cmd: "modprobe bnx2"
  Config Status: cfg=no, avail=yes, need=no, active=unknown
  Attached to: #20 (PCI bridge)
....

Created attachment 742238 [details]
linuxrc-debug level 4 incl. hwinfo.output

As proposed this tar file contains all log files from /var/lib/YaST2 when linuxrc started with debug level 4. 

hwinfo --netcard output also added.

Adding the proposed sequence: 

"debug.wait=net:60 ifcfg=eth0=dhcp4,DHCLIENT_WAIT_AT_BOOT=30"

to the kernel append  remedies the situation. The installer system is successfully started. 

Seems it is indeed a timing problem. 

BR Florian

As it looks like a timing problem related to driver/hardware and there is a solution for it, there is nothing the YaST team can do here. I am closing the bug report by now.

Please, feel free to reopen it you do not agree with the resolution.

And thanks for reporting!