Ralph Shumaker
Tue, 02 Sep 2008 02:07:45 -0700
James G. Sack (jim) wrote:
Ralph Shumaker wrote:James G. Sack (jim) wrote:Ralph Shumaker wrote: .. Did packet counts go down without unload/reload of any wireless modules or rebooting?
Eh, *I* did not write that. :)
Well, obviously rebooting will reset the numbers, so I don't think that's what you're asking. I've been watching via: $ while true; do (dmesg | grep ipw | wc ; dmesg | grep Failed | wc ; dmesg | grep Restarting | wc ; iwconfig eth1 ; ifconfig eth1 ; uptime ; sleep 60); done Every time a new "Failed" error message appears, RX and TX packet counts reset back to 0. So far, no other changes have appeared, but neither have any error messages about a firmware failure. Presently, the uptime is 2:21, with 79 "Failed ..." messages and 0 "... Restarting." messages.Ummm, that doesn't seem right at all (I would say it SB impossible, but..)SB? (Should Be?)Yeah. Aha, each failure seems to cause the module to go through some kind of reset, the same as if you had unloaded and reloaded it (eg via modprobe). Or, maybe network manager is doing that? That sounds like a serious error. You might want to look around on developer lists for that chip? I don't quite know where off the top of my head -- maybe DJA will (has?) point(ed) something out.
I shall await his advice.
On googling your error message I did run across suggestions of putting WIRELESS_AP_SCANMODE='2' in your etc/sysconfig/network-scrips/ifcfg-eth1$ ls /etc/sysconfig/network-scripts/ifcfg-* gives ifcfg-eth0 and ifcfg-lo. Should I copy ifcfg-eth0 to ifcfg-eth1 and just drop the first line (which looks like a commented-out description of eth0) and adjust the DEVICE line and the HWADDR line to fit? Then add that line? $ cat /etc/sysconfig/network-scripts/ifcfg-eth0 # Broadcom Corporation NetXtreme BCM5751 Gigabit Ethernet PCI Express DEVICE=eth0 BOOTPROTO=dhcp HWADDR=xx:xx:xx:xx:xx:xx ONBOOT=yes DHCP_HOSTNAME=localhost.localdomain NM_CONTROLLED=I'm not quite sure. The new shiny "network manager" tool kind of supercedes (and may even ignore) old-style config files, so I'm inclined to go minimal, with something like DEV=eth1 ONBOOT=yes BOOTPROTO=dhcp WIRELESS_AP_SCANMODE='2' maybe the TYPE=WIRELESS NM_CONTROLLED=1 might be worth trying. (rem: no spaces around '=' or at end-of-line)
Whoops. I made the changes I thought necessary from ifcfg-eth0 to make it somewhat more fit for the eth1 interface, but then forgot to include the suggested addition. Making adjustments now. I left in the HWADDR= line since I knew the address to plug in. But to no avail. After adjusting, then rebooting, it comes back up and promptly has 12 of the Failed error messages within the first 3 minutes. Perhaps I should remove the HWADDR after all?
And another suggestion to turn off hardware crypto via a file (create if necessary), etc/modprobe.d/ipw2200, as follows: options ipw2200 hwcrypto=0 (If there are already other options in a line like this, just append the hwcrypto=0 part.The file did not exist. I made it as directed. How do I make it take effect without rebooting?You would have to # modprobe -r ipw2200 # modprobe ipw2200 You may have to disable something (networking) before modprobe will allow removal (then reenable, of course).
Thanks. I'll keep this message for later retrieval.
I really don't understand what the messages really mean, so I'm just guessing. Scan mode has something(?) to do with APs that don't broadcast (or ..), so conceivably that might fit your environment.Hey, sometimes I make guesses too. No worries. Besides, this is a fresh install. He had no problem with me wiping XP away from the laptop he bought from someone. The XP was useless to him because the password was unknown. So I can mess it up and just blow it away and install again.Good luck, ..jim
Thanks. -- A humble foreign policy. No nation building. Don't police the world. --Congressman Ron Paul -- KPLUG-List@kernel-panic.org http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list