wan port dies - or not???

[piri]

hey!

I'm not sure if it's the famous "wan dies" issue but I have the following problem:

- Wl-500g running in Homegateway mode connected to Cable Modem with fixed IP and fixed DNS
- Running the busybox_httpd via firewall rule 8080 -> 80 from outside (in post-boot and post-firewall scripts)
- Running samba (in post-boot script)
- Running on Auto speed (don't know what speed my cable modem is running on)
- upgraded from 1.8.1.7-2a to 1.8.1.7-3 (must admit that the first upgrade failed so I did it a 2nd time and it succeeded then)

After that upgrade my WAN connection doesn't come up at boot time anymore, only pressing the "disconnect/connect" buttons (although it says "conneced" in the link status field) in the webconsole subsequently, connection is being established. After a short time of web browsing (no huge data transfers) WAN connection is lost again, bringing it back up only with the buttons in the web console.

Also tried

et -i eth1 down/up

(thought it's similar to the buttons...)
as well as

et -i eth1 speed 10half/10full/100half/100full

but with no effect, only buttons work.

Tried also the following with no effect:
- installing 1.8.1.7-3 in recovery mode
- resetting to factory defaults via web console
- downgrading to 1.8.1.7-2a
- installing 1.8.2.4
- setting number of max connections to 4096

Also typing

et -i eth1 dump

only brings up a "?" or an empty line, I don't know what that means...

The only thing I noticed is that all trouble begins when flashfs comes into work, i.e. after upgrading I set up the scripts and use flashfs (save, commit, enable, reboot) and after that WAN connection is lost.
Then the post-firewall script is not executed anymore and samba does not work (many smbd processes open up, but isn't this a different issue?...I also have two unknown "ping" processes, but I think I'm getting paranoid )

Sadly

flashfs disable

and rebooting doesn't do the trick either....

Could it be that I messed up flashfs upon the failed upgrade? Can I recover it in any ways or do a total cleanup? Is there a way to monitor when the WAN dies?

Meanwhile I don't know where to look at anymore and what to try next, all I tried didn't bring my any further....

cheers, thomas.

[piri]

I just found out that it also must have something to do with DNS resolution:

- After rebooting only websites that I surfed before reboot seem to work (must be somewhere in a cache), all others don't.
- But pinging them in the wl500g telnet console works i.e WAN connection is indeed established, only the DNS lookup failed.
- Typing the correspoding IP address in the address field of the browser also worked...

... but: after a certain time the WAN finally died, i.e. even IP addresses were not pingable via telnet anymore

Now I'm totally confused....
cheers, thomas.

[Oleg]

Well, flashfs itself does not cause any problems, your scripts could cause problems.
type "flashfs disable" and reboot your router. Check status then.
If it still does not work, post system log and ps ax output here.

[Styno]

I praise your effords to find a solution for your problem yourself.

I don't think you have the 'WAN dies' problem, certainly if you can still ping an IP address (whithout resolving) while you have the problem. You might try to setup a different DNS server for your WL-500g, or try the set it on your Windows box.

[piri]

Thanks for your quick replies guys,

In the meantime I tried the following:

- Resetting the router to defaults, rebooting
- Making a "clean" install of 1.8.1.7-3 via recovery tool
- Resetting and rebooting
- Editing my settings in web console
- Rebooting

Didn't touch the flashfs, still not working, only could get it to life with connect/disconnect buttons...ping isnt working either....

Here's the dmesg output after a reboot:

Dec 31 13:00:01 kernel: klogd started: BusyBox v1.00 (2004.11.14-18:33+0000)
Dec 31 13:00:01 kernel: CPU revision is: 00024000
Dec 31 13:00:01 kernel: Loading BCM4710 MMU routines.
Dec 31 13:00:01 kernel: Primary instruction cache 8kb, linesize 16 bytes (2 ways)
Dec 31 13:00:01 kernel: Primary data cache 4kb, linesize 16 bytes (2 ways)
Dec 31 13:00:01 kernel: Linux version 2.4.20 (root@omnibook) (gcc version 3.2.3 with Broadcom modifications) #3 Wed Nov 10 22:17:25 MSK 2004
Dec 31 13:00:01 kernel: Determined physical RAM map:
Dec 31 13:00:01 kernel:  memory: 01000000 @ 00000000 (usable)
Dec 31 13:00:01 kernel: On node 0 totalpages: 4096
Dec 31 13:00:01 kernel: zone(0): 4096 pages.
Dec 31 13:00:01 kernel: zone(1): 0 pages.
Dec 31 13:00:01 kernel: zone(2): 0 pages.
Dec 31 13:00:01 kernel: Kernel command line: root=/dev/mtdblock2 noinitrd init=/linuxrc console=ttyS0,115200
Dec 31 13:00:01 kernel: CPU: BCM4710 rev 0 at 125 MHz
Dec 31 13:00:01 kernel: !unable to setup serial console!
Dec 31 13:00:01 kernel: Calibrating delay loop... 82.94 BogoMIPS
Dec 31 13:00:01 kernel: Mount-cache hash table entries: 512 (order: 0, 4096 bytes)
Dec 31 13:00:01 kernel: Buffer-cache hash table entries: 1024 (order: 0, 4096 bytes)
Dec 31 13:00:01 kernel: Page-cache hash table entries: 4096 (order: 2, 16384 bytes)
Dec 31 13:00:01 kernel: Checking for 'wait' instruction...  unavailable.
Dec 31 13:00:01 kernel: POSIX conformance testing by UNIFIX
Dec 31 13:00:01 kernel: PCI: Fixing up bus 0
Dec 31 13:00:01 kernel: PCI: Fixing up bridge
Dec 31 13:00:01 kernel: PCI: Fixing up bus 1
Dec 31 13:00:01 kernel: Initializing RT netlink socket
Dec 31 13:00:01 kernel: Starting kswapd
Dec 31 13:00:01 kernel: NTFS driver v1.1.22 [Flags: R/O]
Dec 31 13:00:01 kernel: pty: 256 Unix98 ptys configured
Dec 31 13:00:01 kernel:  Amd/Fujitsu Extended Query Table v1.1 at 0x0040
Dec 31 13:00:01 kernel: Physically mapped flash: Swapping erase regions for broken CFI table.
Dec 31 13:00:01 kernel: number of CFI chips: 1
Dec 31 13:00:01 kernel: Flash device: 0x400000 at 0x1fc00000
Dec 31 13:00:01 kernel: Physically mapped flash: squashfs filesystem found at block 1101
Dec 31 13:00:01 kernel: Creating 5 MTD partitions on "Physically mapped flash":
Dec 31 13:00:01 kernel: 0x00000000-0x00040000 : "pmon"
Dec 31 13:00:01 kernel: 0x00040000-0x003e0000 : "linux"
Dec 31 13:00:01 kernel: 0x001137f4-0x003e0000 : "rootfs"
Dec 31 13:00:01 kernel: 0x003f0000-0x00400000 : "nvram"
Dec 31 13:00:01 kernel: 0x003e0000-0x003f0000 : "config"
Dec 31 13:00:01 kernel: sflash: chipcommon not found
Dec 31 13:00:01 kernel: ip_conntrack version 2.1 (128 buckets, 1024 max) - 344 bytes per conntrack
Dec 31 13:00:01 kernel: ip_conntrack_pptp version 1.9 loaded
Dec 31 13:00:01 kernel: ip_nat_pptp version 1.5 loaded
Dec 31 13:00:01 kernel: ip_tables: (C) 2000-2002 Netfilter core team
Dec 31 13:00:01 kernel: ipt_time loading
Dec 31 13:00:01 kernel: 802.1Q VLAN Support v1.7 Ben Greear <greearb@candelatech.com>
Dec 31 13:00:01 kernel: All bugs added by David S. Miller <davem@redhat.com>
Dec 31 13:00:01 kernel: FAT: bogus logical sector size 33024
Dec 31 13:00:01 kernel: FAT: bogus logical sector size 33024
Dec 31 13:00:01 kernel: NTFS: Unable to set blocksize 512.
Dec 31 13:00:01 kernel: VFS: Mounted root (squashfs filesystem) readonly.
Dec 31 13:00:01 kernel: Warning: unable to open an initial console.
Dec 31 13:00:01 kernel: Algorithmics/MIPS FPU Emulator v1.5
Dec 31 13:00:01 kernel: eth0: Broadcom BCM47xx 10/100 Mbps Ethernet Controller 1.3.2.0
Dec 31 13:00:01 kernel: default: 1a60
Dec 31 13:00:01 kernel: eth1: Broadcom BCM47xx 10/100 Mbps Ethernet Controller 1.3.2.0
Dec 31 13:00:01 kernel: PCI: Enabling device 01:02.0 (0004 -> 0006)
Dec 31 13:00:01 kernel: eth2: Broadcom BCM4320 802.11 Wireless Controller 1.3.2.0
Dec 31 13:00:02 kernel: usb.c: USB device 2 (vend/prod 0x781/0x7104) is not claimed by any active driver.
Dec 31 13:00:04 kernel:   Vendor: Generic   Model: STORAGE DEVICE    Rev: 1033
Dec 31 13:00:04 kernel:   Type:   Direct-Access                      ANSI SCSI revision: 02
Dec 31 13:00:04 kernel: Attached scsi removable disk sda at scsi0, channel 0, id 0, lun 0
Dec 31 13:00:04 kernel: SCSI device sda: 512000 512-byte hdwr sectors (262 MB)
Dec 31 13:00:04 kernel: sda: Write Protect is off
Dec 31 13:00:05 kernel: default: 3a60
Dec 31 13:00:05 kernel: lp driver: get device ID
Dec 31 13:00:05 kernel: neg fail
Dec 31 13:00:07 kernel: Adpt 2 c000
Dec 31 13:00:08 kernel: Adpt 1 c010
Dec 10 06:54:46 kernel: Adpt 0 c010
Dec 10 06:54:47 kernel: Link up c010 8110
Dec 10 06:54:47 kernel: Adpt ffffffff 8110
Dec 10 06:54:48 kernel: lp driver: get device ID
Dec 10 06:54:48 kernel: neg fail
Dec 10 17:54:48 kernel: neg fail
Dec 10 17:54:52 ntp client: time is synchronized to time.nist.gov 
Dec 10 17:54:52 kernel: VFS: Can't find ext3 filesystem on dev sd(8,1).
Dec 10 17:54:52 kernel: MSDOS FS: Using codepage 950
Dec 10 17:54:52 kernel: MSDOS FS: IO charset cp950
Dec 10 17:54:53 USB storage: vfat fs mounted to /tmp/harddisk 
Dec 10 06:54:53 kernel: ext3: No journal on filesystem on sd(8,2)
Dec 10 06:54:53 kernel: MSDOS FS: Using codepage 950
Dec 10 06:54:53 kernel: MSDOS FS: IO charset cp950
Dec 10 06:54:53 kernel: FAT: bogus logical sector size 0
Dec 10 06:54:53 kernel: VFS: Can't find a valid FAT filesystem on dev 08:02.
Dec 10 06:54:53 kernel: FAT: freeing iocharset=cp950
Dec 10 06:54:53 kernel: FAT: bogus logical sector size 0
Dec 10 06:54:53 kernel: VFS: Can't find a valid FAT filesystem on dev 08:02.

And here the ps ax output

  PID  Uid     VmSize Stat Command
    1 piri        576 S   /sbin/init 
    2 piri            SW  [keventd]
    3 piri            RWN [ksoftirqd_CPU0]
    4 piri            SW  [kswapd]
    5 piri            SW  [bdflush]
    6 piri            SW  [kupdated]
    8 piri            SW  [mtdblockd]
   37 piri        296 S   telnetd 
   43 piri        376 S   httpd eth1 
   44 piri        488 S   nas /tmp/nas.lan.conf /tmp/nas.lan.pid lan 
   46 piri        356 S   syslogd -m 0 -O /tmp/syslog.log -S -l 6 
   49 piri        312 S   klogd 
   50 nobody      472 S   dnsmasq 
   51 piri            SW  [khubd]
   61 piri        276 S   lpd 
   65 piri        248 S   p9100d -f /dev/usb/lp0 0 
   66 piri        248 S   p9101d -f /dev/printers/0 1 
   70 piri            SW  [usb-storage-0]
   71 piri            SW  [scsi_eh_0]
   79 piri        316 S   infosvr br0 
   80 piri        460 S   watchdog 
   82 piri        336 S   ntp 
   86 piri        504 S   -sh 
   91 piri        400 S   stupid-ftpd 
   96 piri        364 R   ps ax

@Styno, the DNS Servers are those provided by my Cable ISP, I also have them entered in the IP Config Settings of the router....when connecting WAN directly to my Powerbook with those settings everything works fine....

thanks a lot &
cheers, thomas

[Antiloop]

how do you configure your router?

by putting back your saved configuration file?

[Oleg]

Looks like you've static IP address for connection?
wl500g is working fine, at least it's able to synchronize clocks

Dec 10 17:54:52 ntp client: time is synchronized to time.nist.gov

Do you've any firewall on your PC?

[brubber]

Supposing your computer is not running DHCP or DNS server try this:

On your modem:

Enable DHCP
Enable DNS server

On your router:
WAN IP: Get automically
Get DNS server automatically:Yes
If you have a Speedtouch modem the DNS server will probably be 10.0.0.138 if you need to set it manually

LAN DHCP: Enable
DNS and WINS server: leave blank or use 192.168.1.1 for DNS and leave other fields blank

WLAN DHCP disable Wireless firewall if you have it in your firmware.

On your workstation: Get IP automatically, preferred DNS servers either leave blank or enter 192.168.1.1, renew IP adress

If this doesn't work check you firewall config.

[piri]

First of all thanks a lot guys for all your help!!!

how do you configure your router?

by putting back your saved configuration file?

No, to be on the safe side I entered all settings manually...

Do you've any firewall on your PC?

Only the one from WL-500g, which is enabled by default...

On your modem:

Enable DHCP
Enable DNS server

I have a Cable modem with a static IP which only can be remotely configured by my ISP (chello), who also provides me with DNS servers, which I enter in the WL-500g IP Config along with my static IP address and gateway.

I noticed that a short time after rebooting everything works fine (Maybe when a certain module during bootup hasnt been loaded yet...), then this DNS problem occurs and a while after (say ca. 10 mins) WAN connection finally dies...

cheers, thomas

[piri]

Now I got it: DNS requests from WAN to LAN/WLAN are not processed anymore, i.e. www.chupa.nl is pingable on the router, but not on the client machines (hostname not found).
I tried to restart dnsmasq (because it came closest to those processes I thought they were responsible for DNS resolving, though I really don't know...), but with no effect. Only disconnect/reconnect buttons help.

When I enter one WAN DNS server in the 1st field of the DNS Servers in the DHCP Section of IP Config in the Web Console, everything works fine, except the name of WL-500g (i.e. my.router) is not resolved anymore.

I also have two unknown "ping" processes, but I think I'm getting paranoid

Also my router has now three open ping processes to 140.113.1.1 (ns1.NCTU.edu.tw, National Chiao Tung University), could it be that I have been hacked?????

I'm really disappointed
cheers, thomas.

[piri]

Oleg, can you please explain what services are responsible for DNS and DHCP for LAN/WLAN that I can take a closer look into.

It's still so that after a router reboot a WAN DNS lookup is not possible on the LAN/WLAN clients through the DNS server 192.168.1.1. Only after a disconntect/connect sequence via web console (I wonder what those buttons do, I haven't found a working command line equivalent so far ) it works fine...

Also after pressing those buttons that magic ping processes to 140.113.1.1 come up (I really wonder if thats something to do with above issue), I wish I could find the config file where this IP address is located

thanks a lot for your help so far,
cheers, thomas.

[Oleg]

Please type

cat /etc/dnsmasq.conf

and post results here.
Both DHCP and DNS stuff are handled by dnsmasq.

[piri]

thanks oleg, i suspected that....

here is my dnsmasq.conf:

user=nobody
interface=br0
dhcp-leasefile=/tmp/dnsmasq.log
dhcp-range=lan,192.168.1.2,192.168.1.254,86400
read-ethers
dhcp-authoritative

also took a glimpse into /tmp/dnsmasq.log, where both my external nameservers reside, so no anomalies to the dnsmasq specification, which i found @ www.thekelleys.co.uk

i wonder why dnsmasq is unable to route LAN dns requests from 192.168.1.1 to my external nameservers after a reboot and why the connect/disconnect buttons eliminate that problem...

thanks again for your patience,
cheers, thomas.

[Oleg]

Also, I need this:

cat /etc/resolv.conf
cat /etc/ethers
cat /etc/hosts
iptables -L -v -n

You may try running nslookup on the router:

nslookup www.chupa.nl 127.0.0.1

[piri]

i think i found the reason....

found out that only when I disable the Starcraft(Battle.Net) option in the DMZ properties page of the web console, the problem of not resolving DNS entries between WAN and LAN occurs.

for my understanding and by comparing the iptables output for the option enabled and disabled, all an enabled battle.net setting does is adding the following entries to the iptables chains:

FORWARD
0     0 ACCEPT     udp  --  *      *       0.0.0.0/0            0.0.0.0/0          udp dpt:6112

PREROUTING
0     0 NETMAP     udp  --  *      *       0.0.0.0/0            84.113.4.16        udp spt:6112 192.168.1.0/24

POSTROUTING   
0     0 NETMAP     udp  --  *      *       192.168.1.0/24       0.0.0.0/0          udp dpt:6112 84.113.4.16/32

All other chains keep unchanged.
For my setup that means that not having these entries in the iptables chains all external DNS requests between WAN and LAN are not resolved....very strange....

The files your requested show no anomalies to me though...

/etc/resolv.conf contains the two entries of my nameservers in the form

nameserver <dns ip address1>
nameserver <dns ip address2>

/etc/group contains my statically mapped mac to ip addresses in the form

<macaddr1> 192.168.1.2
<macaddr2> 192.168.1.3
<macaddr3> 192.168.1.4
<macaddr4> 192.168.1.5

/etc/hosts says

127.0.0.1 localhost.localdomain localhost
192.168.1.1 batarillofred my.router my.WL500g

Do you have any explaination for this behaviour?
cheers, thomas.