Forgot to mention I'm currently using Milestone 6. Linux 2.6.31-rc7-4-desktop #1 SMP PREEMPT 2009-08-24 17:40:12 +0200 x86_64 x86_64 x86_64 GNU/Linux

Thank you for the report.
Yes, I can see the error message on Milestone 6 either.

wlan0     Starting DHCP4 client. .                                        
wlan0     DHCP4 client NOT running 

The problem disappear when I switch off the "Enable IPv6" in the Yast2->"network devices"->Network settings". Let us ask the maintainer of the sysconfig package. Marius, you are the maintainer of the rcnetwork script. Can you take over please ?

I'm using Milestone 7 now, with kernel-2.6.31-rc9-7-desktop. Switching off "Enable IPv6" still gives me the 'DHCP4 client NOT running' error for wlan0. 

eth0 is working normally.

(In reply to comment #2)
> Marius, you are the maintainer of the rcnetwork script. Can
> you take over please ?
Sure.

Kevin, Vladimir,
can you take a look into 'dmesg' output and into /var/log/messages,
/var/log/boot.msg if you find kernel Oopses as in bug 537350 comment 6?
[Must not be related - just because you've disabled IPv6.]

(In reply to comment #0)
> Everything I look at shows that wlan0 is up and running except for
> an error that 'dhcp4 client is not running'. 

I'm going now to try to reproduce the dhcp start-report problems.

Further:
 - are there two dhcpcd processes runing after the failure (one for
   eth0 one for wlan0) or only one? See "ps axwww | grep dhcpcd".

When you can clearly reproduce this, please try

ifdown-dhcp wlan0 ; bash -x ifup-dhcp wlan0 > /tmp/ifup-dhcp-wlan0.out

and attach the /tmp/ifup-dhcp-wlan0.out file. It would be helpful.

It looks like some timing issue in checking of the dhcpcd status while
the fork that the dhcpcd client does when it gets an IP...

(In reply to comment #4)

> can you take a look into 'dmesg' output and into /var/log/messages,
> /var/log/boot.msg if you find kernel Oopses as in bug 537350 comment 6?

No Oops here.

But, I have found the problem. It is the new nl80211 extension. If the wpa_supplicant is running with the -Dnl80211 then it does not authenticate. -Dwext works fine.

# ifup wlan0
    wlan0     device: Intel Corporation Wireless WiFi Link 5100
    wlan0     starting wpa_supplicant

It seems now as a wpa-supplicant problem with the new nl80211 extension here. I am not sure if this is the same problem for Kevin. Kevin could you post the output of the "# iwconfig wlan0" and "# ps ax | grep wpa" ?

I will post it tonight as I'm 170 miles away from my linux box right now.

iwconfig wlan0

wlan0     IEEE 802.11bg  ESSID:"2WIRE669"
          Mode:Managed  Frequency:2.412 GHz  Access Point: 00:22:A4:8A:C1:D1
          Bit Rate=54 Mb/s   Tx-Power=27 dBm
          Retry  long limit:7   RTS thr:off   Fragment thr:off
          Encryption key:5D76-D85B-1655-75C6-B97F-1376-0530-0175-2364-4904-6A37-CD78-69F7-8F41-2CC3-1EAA [2]
          Power Management:off
          Link Quality=47/70  Signal level=-63 dBm  Noise level=-94 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:0   Missed beacon:0

ps ax | grep wpa

 3200 ?        Ss     0:00 wpa_supplicant -iwlan0 -c/var/run/wpa_supplicant-wlan0.conf -Dwext -P/var/run/wpa_supplicant/wlan0.pid -B

Based on the info in Comment #9 Kevin got problems despite the fact that he uses the standard wireless extension -Dwext. The AP and encryption key is set. I observe no problems on my systems with -Dwext . Kevin could you provide the information form the Comments #4,5,6 ?

ps axwww | grep dhcpcd

 2327 ?        Ss     0:00 /sbin/dhcpcd --netconfig -L -E -HHH -c /etc/sysconfig/network/scripts/dhcpcd-hook -t 0 -h linux-opsj eth0
 3854 ?        Ss     0:00 /sbin/dhcpcd --netconfig -L -E -G -c /etc/sysconfig/network/scripts/dhcpcd-hook -t 0 -h linux-opsj wlan0
 8982 pts/0    S+     0:00 grep dhcpcd

Created attachment 318600 [details]
dmesg 

Here is dmesg from tonight.

I also noticed tonight that after doing rcnetwork restart, I got a failed messaged on both eth0 and wlan0. But eth0 is working and wlan0 isn't.

Setting up network interfaces:
    eth0      device: nVidia Corporation MCP51 Ethernet Controller (rev a3)
    eth0      Starting DHCP4 client.
    eth0      DHCP4 client NOT running
    eth0                                                             failed
    wlan0     device: Atheros Communications Inc. AR2413 802.11bg NIC (rev 01)
    wlan0     starting wpa_supplicant
    wlan0     Starting DHCP4 client. . .
    wlan0     DHCP4 client NOT running
    wlan0                                                            failed

Created attachment 318601 [details]
wlan0.out

I'm not sure what I did wrong, but wlan0.out attachment only contains one line. I'm going to add another attachment with more details.

Created attachment 318602 [details]
2wlan0.out

I'm using kernel 2.6.31-8-desktop now and I've noticed some changes. Notice the output of iwconfig wlan0 and ps ax | grep wpa in comment #9. Now the output is this:

iwconfig wlan0

wlan0     IEEE 802.11bg  ESSID:""
          Mode:Managed  Frequency:2.412 GHz  Access Point: Not-Associated
          Tx-Power=0 dBm
          Retry  long limit:7   RTS thr:off   Fragment thr:off
          Encryption key:off
          Power Management:off
          Link Quality:0  Signal level:0  Noise level:0
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:0   Missed beacon:0

ps ax | grep wpa
16418 pts/0    R+     0:00 grep wpa


Comparing comment #11 to what I get now:

ps axwww | grep dhcpcd

12515 ?        Ss     0:00 /sbin/dhcpcd --netconfig -L -E -HHH -c /etc/sysconfig/network/scripts/dhcpcd-hook -t 0 -h linux-opsj eth0

As you can see, wlan0 isn't listed.

The result of rcnetwork status:

rcnetwork status 
                                   
Checking optional network interfaces:                                        
    eth0      device: nVidia Corporation MCP51 Ethernet Controller (rev a3)  
    eth0      DHCP4 client (dhcpcd) is running                               
    eth0      IP address: 192.168.1.69/24                                    
    eth0      IP address: 192.168.1.69/24                                    
    eth0                                                             running 
    wlan0     device: Atheros Communications Inc. AR2413 802.11bg NIC (rev 01)
    wlan0     DHCP4  NOT running 
               
cannot find 'iw', please install                                              
    wlan0                                                            dead     
Checking mandatory network interfaces:
    lo
    lo        IP address: 127.0.0.1/8
    secondary lo IP address: 127.0.0.2/8
    lo                                                               running
Checking service network .  .  .  .  .  .  .  .  .  .  .             running

ifstatus wlan0
    wlan0     device: Atheros Communications Inc. AR2413 802.11bg NIC (rev 01)
DHCP4 client NOT running
wlan0 is up
4: wlan0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN qlen 1000
    link/ether 00:24:01:12:a5:27 brd ff:ff:ff:ff:ff:ff
cannot find 'iw', please install


I haven't seen 'cannot find 'iw', please install' before. I also haven't seen 'dead' in red letters before. 

Could this be a problem with the ath5k module?

I just realized I'm no longer getting an ip address for wlan0.

(In reply to comment #17)
     Tx excessive retries:0  Invalid misc:0   Missed beacon:0
 
> cannot find 'iw', please install                                              
>     wlan0                                                            dead     
> I haven't seen 'cannot find 'iw', please install' before. I also haven't seen
> 'dead' in red letters before. 
> Could this be a problem with the ath5k module?

The /usr/sbin/iw comes from the iw package. This package is needed. The iw shall be the "next generation" iwconfig replacement. Try to update to the Milestone 7 first.

I'm currently running Milestone 7 with 2.6.31-8-desktop kernel. The last clean install I did was Milestone 5 and I've been using the factory repos with zypper up to upgrade to the latest Milestones. Looks like iw didn't get installed with Milestone 5 so I installed iw-0.9.15-1.3 (last updated Aug 30) from the factory repo. 

All the errors I mentioned above went away, but I still can't ping any addresses other than the assigned ip address. 

ifstatus wlan0
    wlan0     device: Atheros Communications Inc. AR2413 802.11bg NIC (rev 01)
DHCP4 client (dhcpcd) is running
IP address: 192.168.1.68/24
wlan0 is up
4: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:24:01:12:a5:27 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.68/24 brd 192.168.1.255 scope global wlan0
    inet6 fe80::224:1ff:fe12:a527/64 scope link
       valid_lft forever preferred_lft forever
    wlan0     IP address: 192.168.1.68/24
bssid=00:22:a4:8a:c1:d1
ssid=2WIRE669
id=0
pairwise_cipher=TKIP
group_cipher=TKIP
key_mgmt=WPA-PSK
wpa_state=COMPLETED
ip_address=192.168.1.68

Don't know if this is relevant or not, but I copied the results of 'ifstatus wlan1' from my install of openSUSE 11.1 (where wireless is working) and noticed I have the following lines that aren't in the output I posted in comment #20.

Configured IPv4 routes for interface wlan1:
  169.254.0.0/16 - - wlan1
Active IPv4 routes for interface wlan1:
  192.168.1.0/24  proto kernel  scope link  src 192.168.1.68
  169.254.0.0/16  scope link
  default via 192.168.1.254
1 of 1 configured IPv4 routes for interface wlan1 up

Does this help with trouble shooting or does 11.1 just do things differently?

(In reply to comment #20)
> wpa_state=COMPLETED
> ip_address=192.168.1.68

Maybe the output of the following commands tells more.

# iwconfig wlan0
# netstat -r
# ps ax | grep wpa
# ps ax | grep dhcp

iwconfig wlan0

wlan0     IEEE 802.11bg  ESSID:"2WIRE669"
          Mode:Managed  Frequency:2.412 GHz  Access Point: 00:22:A4:8A:C1:D1
          Bit Rate=24 Mb/s   Tx-Power=20 dBm
          Retry  long limit:7   RTS thr:off   Fragment thr:off
          Encryption key:6FBA-B1D0-009E-99D3-BF4E-7802-6146-4292-EE79-33E6-6266-5CE6-34A4-7B41-F8C5-CC00 [3]
          Power Management:off
          Link Quality=45/70  Signal level=-65 dBm  Noise level=-92 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:0   Missed beacon:0

netstat -r 

Kernel IP routing table
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
192.168.1.0     *               255.255.255.0   U         0 0          0 eth0
192.168.1.0     *               255.255.255.0   U         0 0          0 wlan0
link-local      *               255.255.0.0     U         0 0          0 eth0
loopback        *               255.0.0.0       U         0 0          0 lo
default         home            0.0.0.0         UG        0 0          0 eth0

ps ax | grep wpa

3388 ?        Ss     0:00 wpa_supplicant -iwlan0 -c/var/run/wpa_supplicant-wlan0.conf -Dwext -P/var/run/wpa_supplicant/wlan0.pid -B

ps ax | grep dhcp

2470 ?        Ss     0:00 /sbin/dhcpcd --netconfig -L -E -HHH -c /etc/sysconfig/network/scripts/dhcpcd-hook -t 0 -h linux-opsj eth0
 4431 ?        Ss     0:00 /sbin/dhcpcd --netconfig -L -E -G -c /etc/sysconfig/network/scripts/dhcpcd-hook -t 0 -h linux-opsj wlan0
 8542 pts/0    S+     0:00 grep dhcp

(In reply to comment #23)
> 
> netstat -r 
> 
> Kernel IP routing table
> Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
> 192.168.1.0     *               255.255.255.0   U         0 0          0 eth0
> 192.168.1.0     *               255.255.255.0   U         0 0          0 wlan0
> link-local      *               255.255.0.0     U         0 0          0 eth0
> loopback        *               255.0.0.0       U         0 0          0 lo
> default         home            0.0.0.0         UG        0 0          0 eth0

It seems like a rooting problem. Can you ping the home from both interfaces ?
# ping -I wlan0 home
# ping -I eth0 home

What route does the next command show ?
# ip route get to <ip address of home>

What route does the next command show ?
# ip route get to 209.191.93.52

Can you ping to the outside from home ?

(In reply to comment #24)

> It seems like a rooting problem. Can you ping the home from both interfaces ?
> # ping -I wlan0 home

PING home.gateway.2wire.net (192.168.1.254) from 192.168.1.68 wlan0: 56(84) bytes of data.

^C
--- home.gateway.2wire.net ping statistics ---
178 packets transmitted, 0 received, 100% packet loss, time 176998ms

> # ping -I eth0 home

PING home.gateway.2wire.net (192.168.1.254) from 192.168.1.64 eth0: 56(84) bytes of data.
64 bytes from home (192.168.1.254): icmp_seq=1 ttl=255 time=1.09 ms
64 bytes from home (192.168.1.254): icmp_seq=2 ttl=255 time=0.813 ms
64 bytes from home (192.168.1.254): icmp_seq=3 ttl=255 time=0.991 ms
64 bytes from home (192.168.1.254): icmp_seq=4 ttl=255 time=0.991 ms
64 bytes from home (192.168.1.254): icmp_seq=5 ttl=255 time=1.20 ms
64 bytes from home (192.168.1.254): icmp_seq=6 ttl=255 time=0.971 ms
64 bytes from home (192.168.1.254): icmp_seq=7 ttl=255 time=0.964 ms
64 bytes from home (192.168.1.254): icmp_seq=8 ttl=255 time=0.455 ms
^C
--- home.gateway.2wire.net ping statistics ---
8 packets transmitted, 8 received, 0% packet loss, time 7007ms
rtt min/avg/max/mdev = 0.455/0.936/1.206/0.210 ms

> What route does the next command show ?
> # ip route get to <ip address of home>

192.168.1.254 dev eth0  src 192.168.1.64
    cache  mtu 1500 advmss 1460 hoplimit 64

> What route does the next command show ?
> # ip route get to 209.191.93.52

209.191.93.52 via 192.168.1.254 dev eth0  src 192.168.1.64
    cache  mtu 1500 advmss 1460 hoplimit 64

> Can you ping to the outside from home ?
I don't understand what you want here.

(In reply to comment #25)
> > # ping -I wlan0 home
> 
> PING home.gateway.2wire.net (192.168.1.254) from 192.168.1.68 wlan0: 56(84)
> bytes of data.
> 
> ^C
> --- home.gateway.2wire.net ping statistics ---
> 178 packets transmitted, 0 received, 100% packet loss, time 176998ms

Connect the AP to the home. Did this solve the problem ?

(In reply to comment #26)

> Connect the AP to the home. Did this solve the problem ?

Sorry, but I don't know what you want here. How do I do this?

(In reply to comment #27)
> (In reply to comment #26)
> 
> > Connect the AP to the home. Did this solve the problem ?
> 
> Sorry, but I don't know what you want here. How do I do this?

You can not ping the home from the interface wlan0. Your default gateway is home. You can ping from the interface eth0 via home to the Internet. So I deduce that your AP has no connection to the home.

[wlan0] <---> [AP] <---> [home] <---> (Internet)

Check the connection to the AP.
# ping -I wlan0 <AP IP address>

and check the connection from the AP to the home. Either
# ping -I wlan0 <home IP address>

or connect to the AP's UI and check the diagnostic if available.

How is your AP connected to the home ? It can be an ethernet cable or wireless connection if home got wireless interface.

I have AT&T's U-verse system for internet. The set top box has an ethernet port connected to the wireless router (2WIRE669) with an ethernet cable. I assume this means that my AP is connected to home via ethernet cable. 

I can ping the AP ip address but the home ip address returns 0 received and 100% packet loss like the output of 'ping -I wlan0 home' in comment #25. I don't see any point in diagnosing the AP since my wireless connection is working fine on both my MacBook and openSUSE 11.1. Milestone 7 and 11.1 are installed on the same computer.

(In reply to comment #29)

> I can ping the AP ip address but the home ip address returns 0 received and
> 100% packet loss like the output of 'ping -I wlan0 home' in comment #25. I
> don't see any point in diagnosing the AP since my wireless connection is
> working fine on both my MacBook and openSUSE 11.1. Milestone 7 and 11.1 are
> installed on the same computer.

Weird. You can ping the home via wlan0 from 11.1 but you can not from 11.2. Right ? What can be the difference if the AP is between and you can ping the AP from both 11.1 and 11.2 ?

Anyway, tou can connect to the Internet from eth0. If I understand it right you cannot connect to the Internet from wlan0. It happens when you unplug the ethernet cable from the box. Right ? Post the routing table "netstat -r " then.

(In reply to comment #30)

> Weird. You can ping the home via wlan0 from 11.1 but you can not from 11.2.
> Right ? What can be the difference if the AP is between and you can ping the AP
> from both 11.1 and 11.2 ?

Good question. I don't know.
> Anyway, you can connect to the Internet from eth0. If I understand it right you
> cannot connect to the Internet from wlan0. It happens when you unplug the
> ethernet cable from the box. Right ? 

Yes.

Post the routing table "netstat -r " then.

netstat -r (eth0 unplugged)

Kernel IP routing table       
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
192.168.1.0     *               255.255.255.0   U         0 0          0 eth0 
192.168.1.0     *               255.255.255.0   U         0 0          0 wlan0
link-local      *               255.255.0.0     U         0 0          0 eth0 
loopback        *               255.0.0.0       U         0 0          0 lo   
default         192.168.1.254   0.0.0.0         UG        0 0          0 eth0

netstat -r (eth0 plugged in)

Kernel IP routing table
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
192.168.1.0     *               255.255.255.0   U         0 0          0 eth0
192.168.1.0     *               255.255.255.0   U         0 0          0 wlan0
link-local      *               255.255.0.0     U         0 0          0 eth0
loopback        *               255.0.0.0       U         0 0          0 lo
default         home            0.0.0.0         UG        0 0          0 eth0

I'm still getting the 
'wlan0     DHCP4 client NOT running
    wlan0                                                            failed'
when I do rcnetwork restart but when I do 'ifstatus wlan0' it says wlan0 is up and DHCP4 client (dhcpcd) is running.

(In reply to comment #31)

> netstat -r (eth0 unplugged)
> 
> Kernel IP routing table       
> Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
> 192.168.1.0     *               255.255.255.0   U         0 0          0 eth0 
> 192.168.1.0     *               255.255.255.0   U         0 0          0 wlan0
> link-local      *               255.255.0.0     U         0 0          0 eth0 
> loopback        *               255.0.0.0       U         0 0          0 lo   
> default         192.168.1.254   0.0.0.0         UG        0 0          0 eth0

If the eth0 is unplugged there should be no route via eth0 anymore.

(In reply to comment #32)

> If the eth0 is unplugged there should be no route via eth0 anymore.

With the ethernet cable unplugged, I get the same output from netstat -r after rcnetwork restart and after a reboot as in comment #31 (eth0 unplugged).

I went into YaST Network settings and deleted eth0. Then wlan0 works fine.

netstat -r (eth0 deleted from Network Settings)

Kernel IP routing table
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
192.168.1.0     *               255.255.255.0   U         0 0          0 wlan0
link-local      *               255.255.0.0     U         0 0          0 wlan0
loopback        *               255.0.0.0       U         0 0          0 lo
default         home            0.0.0.0         UG        0 0          0 wlan0

When I reconfigured eth0 with YaST, wlan0 stopped working again.

Sorry,
but it will never work properly to connect to the same network with
two interfaces. Even when there are two default routes, the kernel
will use only first one (one interface).

When you really want to use both at same time and same network config,
you need several custom hooks (http://lartc.org/howto/lartc.rpdb.html).

On every ping you do, you've to use the "-I eth0" or "-I wlan0" option
to force to use one of the interfaces. Another applications (firefox)
don't have any option to bind the outgoing interface.

Please set/add following options ("on cable select" mode in yast):

/etc/sysconfig/network/ifcfg-eth0:
STARTMODE="ifplugd"
IFPLUGD_PRIORITY="10"

/etc/sysconfig/network/ifcfg-wlan0:
STARTMODE="ifplugd"
IFPLUGD_PRIORITY="20"

This causes:
 - eth0 cable connected:
     enable eth0 and disable wlan0.
 - eth0 NOT cable connected:
     disable eth0 and enable wlan0.

Ahm... another way around:

/etc/sysconfig/network/ifcfg-eth0:
STARTMODE="ifplugd"
IFPLUGD_PRIORITY="20"

/etc/sysconfig/network/ifcfg-wlan0:
STARTMODE="ifplugd"
IFPLUGD_PRIORITY="10"

(In reply to comment #35)
> Sorry,
> but it will never work properly to connect to the same network with
> two interfaces. Even when there are two default routes, the kernel
> will use only first one (one interface).
> 
> When you really want to use both at same time and same network config,
> you need several custom hooks (http://lartc.org/howto/lartc.rpdb.html).

I think maybe you misunderstood. I don't want to use both at the same time. I want to use either or. I usually use wlan0 so I don't have to run a 50 ft ethernet cable to the router from my computer. But sometimes I need to use the ethernet cable, such as when I'm changing settings to the wireless router or if I need the added security of a wired connection. 

I also discovered that I can have the same problem in reverse. I did a clean install of Milestone 8 and only wlan0 was configured. Then I added the eth0 configuration and eth0 can't ping any websites, only the assigned ip address. I have never had this problem before, going back to openSUSE 10.0. It only popped up in the Milesone releases for 11.2. 

So I edited /etc/sysconfig/network/ifcfg-wlan0 and eth0 like the examples above and both interfaces are working now. 

Shouldn't YaST do this automatically when more than one interface is configured? Why haven't I had this problem in the past? Is this a problem with ifup but not with NetworkManager?

Surely you don't expect new users to edit /etc/sysconfig/network by hand if they configure both wired and wireless interfaces.

I discovered the 'Activate device' button under the general tab when editing a network interface. I never knew the drop down menu had those six options. 

So this wasn't a bug, just a "I didn't know you could do that" and "I've never had this problem before" kind of thing. 

Thanks for all the help.

(In reply to comment #37)
> I think maybe you misunderstood. I don't want to use both at the same time.

OK.

> So I edited /etc/sysconfig/network/ifcfg-wlan0 and eth0 like the examples above
> and both interfaces are working now.

OK.

> Shouldn't YaST do this automatically when more than one interface is
> configured? Why haven't I had this problem in the past?

YaST did this until now.

> Is this a problem with ifup but not with NetworkManager?

NetworkManager has a daemon that AFAIK automatically selects the fastest
interface, that is, this setup is default for NetworkManager.

> Surely you don't expect new users to edit /etc/sysconfig/network by
> hand if they configure both wired and wireless interfaces.

No, I reassign to the yast2 network maintainer to verify what happens.

ifplugd issue - duplicate of bnc#586376

*** This bug has been marked as a duplicate of bug 586376 ***