[SOLVED-ish] Wifi conx not retained over reboot - in periods...

[CJ]

One of my machines (over periods*) does not automatically connect to my wifi on boot.

* This is not a per-reboot type thing, but it comes in stretches; i.e. first it worked over weeks, then it didn't for weeks; then it did again for weeks again; and currently is does not.

When it fails, I used to, from PCC, continuously removing and setting up again the connection (in PCC). This would always take several attempts, sometimes five, sometimes over 30 times (yes, enough to drive me livid).

I think this has something to do with this machine having had a total of four different cards in it at times: one broke, one turned out not to have native drivers in Linux, one from another machine temporarely when I really couldn't get the connection up, and of course the current card.
(Note only one card at any given time.)

Lately, I have discovered that the following always gets me connected:
Wait for the connection attempts to fail (time out), then modprobe -r rt619ci, then remove the connection in PCC, then set up the connection in PCC.
This leads me to suspect that there might be an issue with the initial loading of the driver.

The card is an older, well-supported ASUS pci card.
Also, I have many good reasons to know that the fault is not with the router (router has no errors, other machines can connect and persist, that this is only in periods with no change to the wifi setup, etc...)

I am on the latest kernel (2.6.38.6-pclos1.bfs), but the behaviour has been seen over older ones too.
I have tried hidding/unhidding SSID - same difference.
Tried using dhcpd instead of dhclient.

I do not know what command output might be helpful, so please ask if you feel you can help!

Cheers!
CJ

[muungwana]

"/var/log/syslog" file most likely have clues of why it is failing when it is failing, look at it when its not working  as expected for clues and post its contents for others to see. A clue is in there.

[CJ]

"/var/log/syslog" file most likely have clues of why it is failing when it is failing, look at it when its not working  as expected for clues and post its contents for others to see. A clue is in there.

muungwana, thank you for helping.
There seems to be a time-out, but as for the reason...?

Code: [Select]

(result on a 'grep wlan' only. If more is needed, let me know.)

May 23 14:08:13 MEDIA klogd: udev: renamed network interface wlan0 to wlan2
May 23 14:08:21 MEDIA klogd: wlan2: deauthenticating from 00:1a:4d:21:82:8c by local choice (reason=3)
May 23 14:08:21 MEDIA klogd: wlan2: authenticate with 00:1a:4d:21:82:8c (try 1)
May 23 14:08:21 MEDIA klogd: wlan2: authenticated
May 23 14:08:21 MEDIA klogd: wlan2: associate with 00:1a:4d:21:82:8c (try 1)
May 23 14:08:21 MEDIA klogd: wlan2: RX AssocResp from 00:1a:4d:21:82:8c (capab=0x431 status=0 aid=2)
May 23 14:08:21 MEDIA klogd: wlan2: associated
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: ifplugd 0.28 initializing.
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: Using interface wlan2/48:5B:39:BE:09:7D with driver <rt61pci> (version: 2.6.38.6-pclos1.bfs)
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: Using detection mode: SIOCETHTOOL
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: Initialization complete, link beat detected.
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: Executing '/etc/ifplugd/ifplugd.action wlan2 up'.
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: client: Determining IP information for wlan2...dhcpcd[2004]: version 5.2.7 starting
May 23 14:08:22 MEDIA dhcpcd[2004]: wlan2: waiting for carrier
May 23 14:08:22 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: wlan2: waiting for carrier
May 23 14:08:23 MEDIA dhcpcd[2004]: wlan2: carrier acquired
May 23 14:08:23 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: wlan2: carrier acquired
May 23 14:08:23 MEDIA dhcpcd[2004]: wlan2: rebinding lease of 192.168.1.104
May 23 14:08:23 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: wlan2: rebinding lease of 192.168.1.104
May 23 14:08:23 MEDIA dhcpcd[2004]: wlan2: carrier lost
May 23 14:08:23 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: wlan2: carrier lost
May 23 14:08:33 MEDIA dhcpcd[2004]: wlan2: broadcasting for a lease
May 23 14:08:33 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: wlan2: broadcasting for a lease
May 23 14:08:52 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: timed out
May 23 14:08:52 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: allowing 8 seconds for IPv4LL timeout
May 23 14:09:00 MEDIA ifplugd(wlan2)[1923]: client: dhcpcd[2004]: timed out
May 23 14:09:00 MEDIA zcip[2833]: interface: wlan2 (48:5B:39:BE:09:7D)
May 23 14:09:00 MEDIA zcip[2833]: Retrieved address 127.255.255.255 for wlan2:9
May 23 14:09:00 MEDIA klogd: device wlan2 entered promiscuous mode
May 23 14:09:08 MEDIA klogd: device wlan2 left promiscuous mode
May 23 14:09:08 MEDIA avahi-daemon[2684]: Joining mDNS multicast group on interface wlan2.IPv4 with address 127.255.255.255.
May 23 14:09:08 MEDIA avahi-daemon[2684]: New relevant interface wlan2.IPv4 for mDNS.
May 23 14:09:08 MEDIA avahi-daemon[2684]: Registering new address record for 127.255.255.255 on wlan2.IPv4.
May 23 14:09:12 MEDIA zcip[2833]: Stored address 127.255.255.255 for wlan2:9
May 23 14:09:12 MEDIA ifplugd(wlan2)[1923]: client:  done.
May 23 14:09:12 MEDIA ifplugd(wlan2)[1923]: Program executed successfully.


CJ

[muungwana]

open the terminal, log in as root and then type these two commands and post their results

/sbin/dhclient wlan2

tail -n 20 /var/log/syslog

Those lines shows your computer can correctly connect to the wireless network and it is asking for an IP address and the network ignores it for some reason.

[CJ]

open the terminal, log in as root and then type these two commands and post their results

/sbin/dhclient wlan2

tail -n 20 /var/log/syslog

Those lines shows your computer can correctly connect to the wireless network and it is asking for an IP address and the network ignores it for some reason.

This should, again, be done when the conx isn't working?
Also, note that lately I set the conx up using dhcpd, not dhclient - should I change that first?

Cheers!
CJ

[muungwana]

yes, try that when the connection isnt working.

you are using "dhcpcd" for a dhcp client while the default one is dhclient, you should mention when you make these kind of changes when reporting problems.

i have never used a system with "dhcpcd" and hence have no idea what to expect from it, i think its best if you go back to the default one(dhclient)

[CJ]

As per my original post:

Tried using dhcpd instead of dhclient.

... both show the same behiviour, but I shall be using dhclient as default from now on to avoid misunderstandings.

More to follow...

CJ

[CJ]

/sbin/dhclient wlan2

This one take its sweet time, then no output.

tail -n 20 /var/log/syslog

Code: [Select]

# tail -n 20 /var/log/syslog
May 23 17:17:29 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 4
May 23 17:17:33 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 10
May 23 17:17:39 MEDIA snmpd[2738]: error on subcontainer 'ia_addr' insert (-1)
May 23 17:17:43 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 9
May 23 17:17:52 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 13
May 23 17:18:09 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 4
May 23 17:18:09 MEDIA snmpd[2738]: error on subcontainer 'ia_addr' insert (-1)
May 23 17:18:13 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 10
May 23 17:18:23 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 19
May 23 17:18:39 MEDIA snmpd[2738]: error on subcontainer 'ia_addr' insert (-1)
May 23 17:18:42 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 19
May 23 17:18:52 MEDIA ntpd_initres[2801]: host name not found EAI_NODATA: 0.pt.pool.ntp.org
May 23 17:18:52 MEDIA ntpd_initres[2801]: host name not found EAI_NODATA: 1.pt.pool.ntp.org
May 23 17:18:52 MEDIA ntpd_initres[2801]: host name not found EAI_NODATA: 2.pt.pool.ntp.org
May 23 17:19:00 MEDIA ntpd[2785]: synchronized to LOCAL(0), stratum 10
May 23 17:19:00 MEDIA ntpd[2785]: kernel time sync status change 0001
May 23 17:19:01 MEDIA dhclient: DHCPDISCOVER on wlan2 to 255.255.255.255 port 67 interval 9
May 23 17:19:09 MEDIA snmpd[2738]: error on subcontainer 'ia_addr' insert (-1)
May 23 17:19:10 MEDIA dhclient: No DHCPOFFERS received.
May 23 17:19:10 MEDIA dhclient: No working leases in persistent database - sleeping.

Something different did happen after this, though:

Code: [Select]

# modprobe -r rt61pci
WARNING: Error removing rt2x00lib (/lib/modules/2.6.38.6-pclos1.bfs/kernel/drivers/net/wireless/rt2x00/rt2x00lib.ko.gz): Resource temporarily unavailable

# modprobe -r rt61pci

After which the conx did come up 'by itself'. Normally, I get no error on the first 'modprobe -r rt61pci', and I have to manually set-up the conx in PCC to get it working.

CJ

[muungwana]

you have a strange problem, your log files says the interface can successfully connect to the wireless network and is asking for an IP address but the network is the one that is not answering it.

The interface working later on after you remove the module suggest the problem is the module in use. What wireless card are you using?

The interface working as expected after you remove that  module suggests the problem is with the module. May be there are two modules, one works and the other doesnt and the one that doesnt takes precedence over the other and it the interface starts to work and this one is removed allowing the other to take control of the card.

you should open "/var/log/syslog" text file and monitor its contents as you try different things. A clue to what works is somewhere in there. You must open the text file with a text editor running with root's privileges

[CJ]

Strange indeed. I have to agree with you as I started out troubleshooting the network. When it fails to obtain an IP address, the router does not even log any incoming attempt to connect... no, there seems little doubt that the issue is with this particular machine.

The card is an ASUS PCI-G31.

I even briefly suspected that it might be a case of loading the rt61pci module as default, and after removing that, a completely different (and better) driver might be loaded. Alas, the hardware centre shows the rt61pci driver both before and after setting it up again, so no go on that score.

Is what you describe - i.e. having to instances of the same module running at the same time - even possible, and if so, do you know how I would go about establishing that?

I shall do as you say (monitor syslog for live changes), but I am not too sure what to try any more.

Thank you very much for your help, much appreciated.
CJ

[muungwana]

You can not have more than one instance or more than one driver loaded to drive a piece of hardware. A module registers itself when it is loaded saying what hardware is responsible for and all other modules that want to claim the same hardware are automatically blocked.

what does the log file say right after it fails to connect and you do the "modprobe -r rt619ci" and then it connects.

There should be an entry after you run the modprobe command

[CJ]

muungwana, I have inserted the excerpt from syslog after removing the driver and remvoing the conx, i.e. from when I set up the conx in PCC.
Apart from all working on try 1, only one things jumps out at me: the changing of regulatory domain to JP (from PT). I seem to recall having seen this before.
Before you suggest it, I have tried changing to JP as the default, and apart from the fact that I am not in JP, it makes no difference.

Code: [Select]

May 24 11:34:03 MEDIA drakconnect[2682]: ### Program is starting ###
May 24 11:34:06 MEDIA drakconnect[2682]: running: /bin/rpm -q --qf %{name}  wireless-tools
May 24 11:34:08 MEDIA drakconnect[2682]: running: dmidecode
May 24 11:34:08 MEDIA drakconnect[2682]: Found settings for driver "rt61pci" in category "network::connection::wireless"
May 24 11:34:08 MEDIA drakconnect[2682]: No kernel_module package for module "rt61pci" is required, skipping
May 24 11:34:08 MEDIA drakconnect[2682]: No tools package for module "rt61pci" is required, skipping
May 24 11:34:08 MEDIA drakconnect[2682]: Required firmware package for module "rt61pci" is already installed, skipping
May 24 11:34:08 MEDIA drakconnect[2682]: Loading module rt61pci
May 24 11:34:08 MEDIA drakconnect[2682]: Settings for driver "rt61pci" applied
May 24 11:34:30 MEDIA drakconnect[2682]: running: /bin/rpm -q --qf %{name}  dhcp-client
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network-scripts/wireless.d/00:1a:4d:21:82:8c
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network-scripts/wireless.d/00:1a:4d:21:82:8c
May 24 11:34:31 MEDIA drakconnect[2682]: changed mode of /etc/sysconfig/network-scripts/wireless.d/00:1a:4d:21:82:8c to 700
May 24 11:34:31 MEDIA drakconnect[2682]: written  interface configuration in /etc/sysconfig/network-scripts/wireless.d/00:1a:4d:21:82:8c
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network-scripts/wireless.d/doghouse
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network-scripts/wireless.d/doghouse
May 24 11:34:31 MEDIA drakconnect[2682]: changed mode of /etc/sysconfig/network-scripts/wireless.d/doghouse to 700
May 24 11:34:31 MEDIA drakconnect[2682]: written  interface configuration in /etc/sysconfig/network-scripts/wireless.d/doghouse
May 24 11:34:31 MEDIA drakconnect[2682]: created file /etc/wpa_supplicant.conf
May 24 11:34:31 MEDIA drakconnect[2682]: running: /usr/sbin/wpa_cli reconfigure
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network-scripts/ifcfg-wlan2
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network-scripts/ifcfg-wlan2
May 24 11:34:31 MEDIA drakconnect[2682]: changed mode of /etc/sysconfig/network-scripts/ifcfg-wlan2 to 700
May 24 11:34:31 MEDIA drakconnect[2682]: written wlan2 interface configuration in /etc/sysconfig/network-scripts/ifcfg-wlan2
May 24 11:34:31 MEDIA drakconnect[2682]: writing host information to /etc/hosts
May 24 11:34:31 MEDIA drakconnect[2682]: created file /etc/hosts
May 24 11:34:31 MEDIA drakconnect[2682]: writing host information to /etc/hosts
May 24 11:34:31 MEDIA drakconnect[2682]: created file /etc/hosts
May 24 11:34:31 MEDIA drakconnect[2682]: set sethostname to MEDIA
May 24 11:34:31 MEDIA drakconnect[2682]: running: /usr/bin/run-parts --arg MEDIA /etc/sysconfig/network-scripts/hostname.d
May 24 11:34:31 MEDIA drakconnect[2682]: modified file /etc/sysconfig/network
May 24 11:34:31 MEDIA drakconnect[2682]: running: /sbin/ifdown wlan2 daemon
May 24 11:34:32 MEDIA ifplugd(wlan2)[2475]: Exiting.
May 24 11:34:32 MEDIA drakconnect[2682]: running: /sbin/ifup wlan2 daemon
May 24 11:34:32 MEDIA klogd: NET: Registered protocol family 17
May 24 11:34:34 MEDIA ifplugd(wlan2)[2876]: ifplugd 0.28 initializing.
May 24 11:34:34 MEDIA ifplugd(wlan2)[2876]: Using interface wlan2/48:5B:39:BE:09:7D with driver <rt61pci> (version: 2.6.38.6-pclos1.bfs)
May 24 11:34:34 MEDIA ifplugd(wlan2)[2876]: Using detection mode: SIOCETHTOOL
May 24 11:34:34 MEDIA ifplugd(wlan2)[2876]: Initialization complete, link beat not detected.
May 24 11:34:34 MEDIA drakconnect[2682]: running: /etc/rc.d/init.d/network-up start
May 24 11:34:34 MEDIA klogd: wlan2: deauthenticating from 00:1a:4d:21:82:8c by local choice (reason=3)
May 24 11:34:34 MEDIA klogd: wlan2: authenticate with 00:1a:4d:21:82:8c (try 1)
May 24 11:34:34 MEDIA klogd: wlan2: authenticated
May 24 11:34:34 MEDIA klogd: wlan2: associate with 00:1a:4d:21:82:8c (try 1)
May 24 11:34:34 MEDIA klogd: wlan2: RX AssocResp from 00:1a:4d:21:82:8c (capab=0x431 status=0 aid=1)
May 24 11:34:34 MEDIA klogd: wlan2: associated
May 24 11:34:34 MEDIA klogd: cfg80211: Calling CRDA for country: JP
May 24 11:34:34 MEDIA klogd: cfg80211: Regulatory domain changed to country: JP
May 24 11:34:34 MEDIA klogd: cfg80211:     (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
May 24 11:34:34 MEDIA klogd: cfg80211:     (2402000 KHz - 2472000 KHz @ 40000 KHz), (N/A, 2000 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (2457000 KHz - 2482000 KHz @ 20000 KHz), (N/A, 2000 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (2474000 KHz - 2494000 KHz @ 20000 KHz), (N/A, 2000 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (4910000 KHz - 4930000 KHz @ 10000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (4910000 KHz - 4990000 KHz @ 40000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (4930000 KHz - 4950000 KHz @ 10000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (5030000 KHz - 5045000 KHz @ 10000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (5030000 KHz - 5090000 KHz @ 40000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (5050000 KHz - 5060000 KHz @ 10000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (5170000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (5250000 KHz - 5330000 KHz @ 40000 KHz), (N/A, 2000 mBm)
May 24 11:34:34 MEDIA klogd: cfg80211:     (5490000 KHz - 5710000 KHz @ 40000 KHz), (N/A, 2300 mBm)
May 24 11:34:34 MEDIA klogd: padlock_aes: VIA PadLock not detected.
May 24 11:34:34 MEDIA klogd: Intel AES-NI instructions are not detected.
May 24 11:34:35 MEDIA ifplugd(wlan2)[2876]: Link beat detected.
May 24 11:34:36 MEDIA ifplugd(wlan2)[2876]: Executing '/etc/ifplugd/ifplugd.action wlan2 up'.
May 24 11:34:37 MEDIA dhclient: DHCPREQUEST on wlan2 to 255.255.255.255 port 67
May 24 11:34:37 MEDIA dhclient: DHCPACK from 192.168.1.1
May 24 11:34:37 MEDIA klogd: martian source 255.255.255.255 from 192.168.1.1, on dev wlan2
May 24 11:34:37 MEDIA klogd: ll header: ff:ff:ff:ff:ff:ff:00:50:7f:b8:ec:08:08:00
May 24 11:34:37 MEDIA avahi-daemon[1702]: Joining mDNS multicast group on interface wlan2.IPv4 with address 192.168.1.104.
May 24 11:34:37 MEDIA avahi-daemon[1702]: New relevant interface wlan2.IPv4 for mDNS.
May 24 11:34:37 MEDIA avahi-daemon[1702]: Registering new address record for 192.168.1.104 on wlan2.IPv4.
May 24 11:34:37 MEDIA NET[3107]: /sbin/dhclient-script : updated /etc/resolv.conf
May 24 11:34:37 MEDIA dhclient: bound to 192.168.1.104 -- renewal in 110150 seconds.
May 24 11:34:37 MEDIA ifplugd(wlan2)[2876]: client: Determining IP information for wlan2... done.
May 24 11:34:38 MEDIA ifplugd(wlan2)[2876]: Program executed successfully.
May 24 11:34:43 MEDIA drakconnect[2682]: modified file /etc/modprobe.conf
May 24 11:34:43 MEDIA drakconnect[2682]: to put in /etc/modprobe.preload evdev
May 24 11:34:43 MEDIA drakconnect[2682]: modified file /etc/modprobe.preload
May 24 11:34:43 MEDIA drakconnect[2682]: ### Program is exiting ###


I need to make mention of another thing before we go any further. Maybe I misread your suggestion about trying things, but...
I tried installing an earlier kernel module (.37.4), then removing the current one (38.6). Starting up with 37 basically did the same thing.
I then reinstalled the current kernel, but no difference - except I notice that starting udev seems to hang for a few seconds in the start-up process..
(The thinking was that, if the rt61pci kernel module was somehow damaged, this might fix it.)

Cheers!
CJ

[AS]

Hi CJ,

apparently there are two driver available for your card (ralink rt61 chipset): rt61pci (also called Legacy), and a new rt2x00pci
http://rt2x00.serialmonkey.com/wiki/index.php/Main_Page

the rt2x00pci is available with PCLinuxOS kernel 2.6.33.7 and with kernel 2.6.38.8 (I have this two installed on my system), so I think is available with any kernel from 2.6.33.7 up the the current.

Worth to try if the new driver may work better, then may be you need to blacklist the rt61pci and manually load the rt2x00pci.
If something is not clear please ask.

AS

[Hootiegibbon]

Cj,

Regarding the 3 previous cards, did you ever remove the config for those cards from PCC?

ie PCC > Network&internet > remove.....

Jase

[mikkl]

This sounds like a driver conflict. You might want to try blacklisting all wireless modules and then enabling only one for installation at boot.

mikkl