rt2500 card not getting dhcp lease

Bug #51081 reported by Ooblick
4
Affects Status Importance Assigned to Milestone
dhcp3 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: dhcdbd

I currently have a pcmcia wireless card with the rt2500 chipset:

# lspci | grep RT2500
0000:02:00.0 Network controller: RaLink Ralink RT2500 802.11 Carbus Reference Card (rev 01)

Local wireless access points are show in network manager. From network manager an access point is selected, one with no WEP.

Expected result:

Network connection established

Actual result:

No connection established. Network manager sits spinning for some time.

This seems to be caused by dhcdbd not invoking dhclient correctly. The following line is in /var/log/messages:

Jun 27 06:34:32 berry dhcbdb: message_handler: message handler not found under /com/redhat/dhcp/ra0 for sub-path ra0.dbus.get.reason

The rt2500 makes the wireless card available as ra0 (not eth1 or others). Manually running "dhclient ra0" successfully obtains a wireless lease and enables connectivity.

This bug is to fix dhcdbd so it successfully invokes dhclient for ra0. It is hoped this will enable network-manager to connect.

Version effected: dapper 6.06

Revision history for this message
Ooblick (txe-stc) wrote : lspci -vv

lspci -vv output for this system

Revision history for this message
Kristoffer Bergström (kabtoffe) wrote :

I get this too! Same log messages and "symptoms".

Revision history for this message
Scott James Remnant (Canonical) (canonical-scott) wrote :

I don't think this is dhcdbd, could you try running this:

        /usr/bin/dbus-send --system \
                           --dest=com.redhat.dhcp --type=method_call \
                           --print-reply --reply-timeout=20000 \
                           /com/redhat/dhcp/ra0 \
                           com.redhat.dhcp.up

Changed in dhcdbd:
status: Unconfirmed → Needs Info
Changed in dhcdbd:
assignee: nobody → keybuk
Revision history for this message
Ooblick (txe-stc) wrote :

Looks like you're right. Running that gives the response:

method return sender=:1.1 -> dest=:1.20
 uint32 0

However looking at what runs before and after with 'ps -ef | grep dbus' I can see that a new process begins running after running your command:

/sbin/dhclient -1 -lf /var/lib/dhcp3/dhclient.ra0.leases -pf /var/run/dhclient.ra0.pid -q -e dhc_dbus=0 -d ra0

This seems to just sit around blocking - killing it and running it causes it to sit on the command line without an IP address being allocated to ra0.

The file /var/lib/dhcp3/dhclient.ra0.leases looks like it has correct data for a lease - however its not configuring the interface with it.

Revision history for this message
Scott James Remnant (Canonical) (canonical-scott) wrote :

Note that this is almost certainly a kernel bug ... if you "ifconfig ra0 up" before running the dhclient command (or dhcdbd poke) does it work?

Revision history for this message
Nafallo Bjälevik (nafallo) wrote :

The device is already up before it can scan.

Changed in dhcp3:
assignee: keybuk → nobody
Revision history for this message
Peter Jochum (peter-jochum) wrote :

Same problem here (Edgy). The network manager sets the essid fine, however aquiring a dhcp-lease fails. If you use dhclient afterwards everything works fine.

I renamed the ra0 device to eth1 in /etc/iftab, but this has no effect either.

Revision history for this message
Launchpad Janitor (janitor) wrote :

[Expired for dhcp3 (Ubuntu) because there has been no activity for 60 days.]

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Bug attachments

Remote bug watches

Bug watches keep track of this bug in other bug trackers.