Windoze doesn't drop WLAN but FC6 does...

[mike-g2]

Hi all,

When I'm running linux my laptop seems to always lose its connection to my wireless router but never seems to in Windoze.

More specifically, I'll often get "Host Unreachable" errors in firefox and/or lose my ssh connections. When this happens I'll fire up the network-config GUI and the connection is still viewed as 'active'. However, if I try pinging my router or cable modem I cannot reach either.

When this happens I'll first try deactivating and then reactivating my connection (which the network-config GUI does w/o errors) but only about 30% of the time will I truly reestablish my connection to the router. Most times it doesn't change anything and I have to go and unplug/replug the router.

This problem NEVER happens in windoze and it's driving me nuts.

I'm running a ThinkPadT41 with built in intel wireless hardware (ipw2100?). I've moved from FC4 to FC6 (a clean install actually, not an upgrade) but still run into the problem.

I'd be happy to provide additional info to do the troubleshoot, but not sure what's relavent.

Please don't let me have to use Windoze. Please help!

Thanks!

Mike

[ppesci]

May be you must to see in /var/log/messages.

Try:

less /var/log/messages

and see if there are a messages from the wireless connection

[mike-g2]

Hi Pietro,

I've looked at the /var/log/messages file in the past and it's never seemed to give much info. It shows me opening the connection, it being made and then, because I can't reach anywhere, being reset. Here's an example snippet:


Jan 28 16:34:50 localhost kernel: ADDRCONF(NETDEV_UP): eth1: link is not ready
Jan 28 16:34:51 localhost kernel: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
Jan 28 16:34:51 localhost dhclient: DHCPREQUEST on eth1 to 255.255.255.255 port 67
Jan 28 16:34:51 localhost dhclient: DHCPACK from 192.168.0.1
Jan 28 16:34:51 localhost avahi-daemon[2340]: New relevant interface eth1.IPv4 for mDNS.
Jan 28 16:34:51 localhost avahi-daemon[2340]: Joining mDNS multicast group on interface eth1.IPv4 with address 192.168.0.2.
Jan 28 16:34:51 localhost avahi-daemon[2340]: Registering new address record for 192.168.0.2 on eth1.
Jan 28 16:34:51 localhost NET[3107]: /sbin/dhclient-script : updated /etc/resolv.conf
Jan 28 16:34:51 localhost dhclient: bound to 192.168.0.2 -- renewal in 113876 seconds.
Jan 28 16:34:52 localhost avahi-daemon[2340]: New relevant interface eth1.IPv6 for mDNS.
Jan 28 16:34:52 localhost avahi-daemon[2340]: Joining mDNS multicast group on interface
eth1.IPv6 with address fe80::20c:f1ff:fe19:7cf9.
Jan 28 16:34:52 localhost avahi-daemon[2340]: Registering new address record for fe80::20c:f1ff:fe19:7cf9 on eth1.
Jan 28 16:49:02 localhost NET[4383]: /sbin/dhclient-script : updated /etc/resolv.conf
Jan 28 16:49:02 localhost avahi-daemon[2340]: Withdrawing address record for 192.168.0.2 on eth1.
Jan 28 16:49:02 localhost avahi-daemon[2340]: Leaving mDNS multicast group on interface eth1.IPv4 with address 192.168.0.2.
Jan 28 16:49:02 localhost avahi-daemon[2340]: iface.c: interface_mdns_mcast_join() called but no local address available.
Jan 28 16:49:02 localhost avahi-daemon[2340]: Interface eth1.IPv4 no longer relevant for mDNS.
Jan 28 16:49:02 localhost avahi-daemon[2340]: Withdrawing address record for fe80::20c:f1ff:fe19:7cf9 on eth1.
Jan 28 16:49:02 localhost avahi-daemon[2340]: Leaving mDNS multicast group on interface eth1.IPv6 with address fe80::20c:f1ff:fe19:7cf9.
Jan 28 16:49:02 localhost avahi-daemon[2340]: iface.c: interface_mdns_mcast_join() called but no local address available.
Jan 28 16:49:02 localhost avahi-daemon[2340]: Interface eth1.IPv6 no longer relevant for mDNS.


Do you see anything relavent? ALso, I know it's possible to have other levels of messaging that might give more info but I'm not sure what program I should have produce the output.

Mike

[ppesci]

Your log appear to be pretty normal, but try to disable ipv6. To do this edit the file /etc/sysconfig/network and change the line:

NETWORKING_IPV6=yes

to

NETWORKING_IPV6=no

and restart the network with:

/etc/init.d/network restart

HTH

[mike-g2]

Pietro,

I have this problem regardless of the ipv6 setting, but I'll note that it is (and was) set to off in the config file. Which makes me wonder why I am getting messages like:

Jan 28 16:49:02 localhost avahi-daemon[2340]: Interface eth1.IPv6 no longer relevant for mDNS.

Could the fact that I've got ipv6 off but am getting this message be a clue to where the problem lies?

Mike

[mwette]

you may be able to "fix" the ipv6 issue by disableing in /etc/avahi/avahi-daemon.conf
I doubt if this is the cause of your wireless issue.

[mike-g2]

Hi Mwette,

Thanks for the note on modifying /etc/avahi/avahi-daemon.conf.

Nevertheless, I am still hoping to get the problem fixed, or at least diagnose it. Losing my connection every 5 mins is almost enough to make me use windoze.

Mike

[ppesci]

Post here if you have good results or not

[Redbear]

Mike,

You are not alone. I just installed Fedora last week - and implemented ndiswrapper - my wireless connection lasts for 60-90 seconds before going out. I can deactivate and activate and get another 60-90 seconds. Please let me know if you find a solution.

I'll do likewise.

[wwagner]

Please check if you find a message "disabling IRQ (number)" in your dmesg log. If so, please check if the IRQ is the one of your network driver (mostly eth0)
In my case using the kernel argument "acpi=off" might help. The other side is that all power management is switched off too.

[mike-g2]

Hi All,

I appreciate the input and suggestions.


Please check if you find a message "disabling IRQ (number)" in your dmesg log. If so, please check if the IRQ is the one of your network driver (mostly eth0)
In my case using the kernel argument "acpi=off" might help. The other side is that all power management is switched off too.

wwagner, I have searched the file but don't find any messages like that. The only time I find an IRQ line is when I hibernate or wake the machine.


you may be able to "fix" the ipv6 issue by disableing in /etc/avahi/avahi-daemon.conf
I doubt if this is the cause of your wireless issue.


Mwette, I changed the avahi-config file, reloaded the file with the debug flag, but that doesnt seem to solve the problem. I am still losing the network connection frequently and, strangely, my /var/log/messages still has ipv6 errors. The var/log/messages is silent when the connection is lost, but when I can reestablish it I get the following output


.
.
.
Jan 29 22:49:17 localhost NET[22690]: /sbin/dhclient-script : updated /etc/resolv.conf
Jan 29 22:49:17 localhost avahi-daemon[2340]: Withdrawing address record for 192.168.0.2 on eth1.
Jan 29 22:49:17 localhost avahi-daemon[2340]: Leaving mDNS multicast group on interface eth1.IPv4 with address 192.168.0.2.
Jan 29 22:49:17 localhost avahi-daemon[2340]: iface.c: interface_mdns_mcast_join() called but no local address available.
Jan 29 22:49:17 localhost avahi-daemon[2340]: Interface eth1.IPv4 no longer relevant for mDNS.
Jan 29 22:49:17 localhost avahi-daemon[2340]: Withdrawing address record for fe80::20c:f1ff:fe19:7cf9 on eth1.
Jan 29 22:49:17 localhost avahi-daemon[2340]: Leaving mDNS multicast group on interface eth1.IPv6 with address fe80::20c:f1ff:fe19:7cf9.
Jan 29 22:49:17 localhost avahi-daemon[2340]: iface.c: interface_mdns_mcast_join() called but no local address available.
Jan 29 22:49:17 localhost avahi-daemon[2340]: Interface eth1.IPv6 no longer relevant for mDNS.
Jan 29 22:49:20 localhost kernel: ADDRCONF(NETDEV_UP): eth1: link is not ready
Jan 29 22:49:21 localhost kernel: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
Jan 29 22:49:21 localhost dhclient: DHCPREQUEST on eth1 to 255.255.255.255 port 67
Jan 29 22:49:21 localhost dhclient: DHCPACK from 192.168.0.1
Jan 29 22:49:21 localhost avahi-daemon[2340]: New relevant interface eth1.IPv4 for mDNS.
Jan 29 22:49:21 localhost avahi-daemon[2340]: Joining mDNS multicast group on interface eth1.IPv4 with address 192.168.0.2.
Jan 29 22:49:21 localhost avahi-daemon[2340]: Registering new address record for 192.168.0.2 on eth1.
Jan 29 22:49:21 localhost NET[22889]: /sbin/dhclient-script : updated /etc/resolv.conf
Jan 29 22:49:21 localhost dhclient: bound to 192.168.0.2 -- renewal in 106097 seconds.
Jan 29 22:49:23 localhost avahi-daemon[2340]: New relevant interface eth1.IPv6 for mDNS.
Jan 29 22:49:23 localhost avahi-daemon[2340]: Joining mDNS multicast group on interface eth1.IPv6 with address fe80::20c:f1ff:fe19:7cf9.
Jan 29 22:49:23 localhost avahi-daemon[2340]: Registering new address record for fe80::20c:f1ff:fe19:7cf9 on eth1.


I'm hoping, but not holding my breath, that this could shine any light on the situation.

Mike

[Redbear]

GOT IT!!!!

Mike,

Try this:

in a terminal window type

su -c "yum -d 0 list kernel"

if you see "kernel.i586 Installed" - that's the issue - at least it was mine.

You should have kernel.i686 installed.

Do this

1) yum remove kernel kernel-devel
2) yum install kernel.i686 kernel-devel.i686
3) yum install (the list of a dozen odd i386 kernel depencies removed in step
#1, such as "gnome-volume-manager" and "compiz")

You will need to be able to view the kernel dependicies that get removed in step 1 - I used two terminal windows - so I could view one and work in the other.

All this orinigally came from the following website.

http://fedoramobile.org/fc-wireless/ndis-yum-livna/


My card started working immediately upon activation - and is still up.

Hope this helps.

[Redbear]

WAIT !!!

GRUB gets corrupted - unable to reboot.

[ppesci]

Reabear:

I more thing is there are incompatibilities between ndiswrapper and the last kernel (2.6.19) ecause a had problems to install the module. If you are using this kernel, try to use your 2.6.18 instead and may be you will need to reinstall ndiwrapper

[mike-g2]

GOT IT!!!!

Mike,

Try this:

in a terminal window type

su -c "yum -d 0 list kernel"

if you see "kernel.i586 Installed" - that's the issue - at least it was mine.

You should have kernel.i686 installed.

Do this

1) yum remove kernel kernel-devel
2) yum install kernel.i686 kernel-devel.i686
3) yum install (the list of a dozen odd i386 kernel depencies removed in step
#1, such as "gnome-volume-manager" and "compiz")

You will need to be able to view the kernel dependicies that get removed in step 1 - I used two terminal windows - so I could view one and work in the other.

All this orinigally came from the following website.

http://fedoramobile.org/fc-wireless/ndis-yum-livna/


My card started working immediately upon activation - and is still up.

Hope this helps.
Hi Redbear,

Yikes! I hope you got your machine runnin' again.

I didn't know about the yum command 'list'. (or setting the debug level, but now I do. Thanks!)

Alas, when I run it, it tells me that I'm running the i686 kernel, as you would hope/expect. So that's apparently not the issue.

Indeed I don't appear to be using/have installed ndiswrapper. (neither yum nor rpm nor locate return anything)

Not sure if I should be happy or disappointed. Still have the same problem, but I didn't have to mess with installing a different kernel type and the dangers that involves.

Anyone else got any ideas?

Mike

[Redbear]

Whew!
I'm back up and typing to you wirelessly. ppesci appears to be correct. I could not get ndiswrapper to work with i686 kernel. I ended up reinstalling fc6 from scratch - letting it default to kernel.i586. I let all of the patches load (227 of them), and did the GCC compatability mod from http://www.mjmwired.net/resources/mjm-fedora-fc6.html.
Then installed ndiswrapper and voila - I'm up. I cannot tell you which part gave me success here. I had done essentially the same thing the first time - except I most likely did more trial and error activities on my first try. It is very possible that I had inadvertently corrupted some necessary file. I don't know if that could be your issue. I was very brave and trusting of patches and mods, however I now think I'm cured of that tendency.
Redbear

[mike-g2]

So I can now often go an hour or so without losing my connection, which is a great improvement. Perhaps modifying /etc/avahi/avahi-daemon.conf ultimately did the trick.

Thanks for all of your input.

Mike

[tsorvoja]

I had the same problem with three Acer Aspire 3102WLMi lapstops. I think I got it working by disabling the wpa_supplicant from services. Now the wpa_supplicant is started only by NetworkManager, when needed.