[Soekris] Soekris] 4801 thats stopped responding

william estrada MrUmunhum at popdial.com
Tue Mar 27 20:21:04 UTC 2007


Tuc,

   Good feed back.  Some options:

1) Is a WiFi link practical?  With a link and another 4801, you could
    build a 'remote operator'.  This would be a great project and I
    would be happy to help you with it.

5) USB HUB??

   Ok, now for debugging options:

1) I still like using an 'active monitor' on your boot console.  This would
    require an output device like a printer or another computer to log the
    messages.  Although it may not tell you what when wrong exactly, it could
    tell you what happened just before the bad thing?  It sure looks like a
    software problem.  It sounds like a very loaded system.  If your system
    is doing a 'Panic', you will be able to see that.

2) Use the Soekris Watchdog function.  Although this will not tell want when
    wrong, it will reset the unit.  By adding an SNMP cold start trap, you will
    know when the system reboots.

3) Is this the only system that is having this problem?


William Estrada
MrUmunhum at popdial.com
Mt-Umunhum-Wireless.net ( 64.124.13.3 )



Tuc at T-B-O-H.NET wrote:
> Hi William,
> 
> 	Thanks.... In reply to your message :
> 
> 1) I can imagine. I'm a repairpersons nightmare. If I had a dime for 
> every "That situation is impossible/I've never heard of that before/
> Thats never broken on me before/etc"... 
> 
> 2) Network isn't running. On the switch it was showing up though.
> SSH wouldn't respond, nor ping, nor any other service I had running
> on the box via TCP (Web, NOCOL, Squid, Telnet{Yea, I know, I know...},
> HTTPS, Sendmail) or UDP (OpenVPN, Bind). 
> 
> 3) OpenVPN has a heartbeat which is how I knew it had gone away.
> 
> 4) 2nd serial is already set up for UPS.
> 
> 5) USB is already in use for Squid cache.
> 
> 6) See 5...
> 
> 	The bigger issue is this is at a remote site that takes 
> 11 hours and a 4WD vehicle to get to... I just happen to be at it
> this week, so I could have done alot of debug and if crashed I
> could get to it easier.
> 
> 		Thanks, Tuc
>> Tuc,
>>
>>    There are several 'things' you can do to help figure out what is broken.
>>
>> 1) First I don't think it is a hardware problem.  The 4801 is a pretty
>>     solid unit.  Mine had been running 24x7 for almost 3 years now.
>>
>> 2) Since it looks like you network is running, maybe?  You should install
>>     SSHD and SNMPD. SSHD will allow you external access.  SNMPD will allow
>>     you external control.  I can help you set up the SNMP MIB if you like.
>>     The flashing network light does not mean that you system is up, it only
>>     means what there is network traffic for other sources.
>>     Can you ping the 4801 at all?
>>
>> 3) I would set up an SNMP watchdog like script that would send out a UDP
>>     SNMP trap say every 60 seconds and watch on another computer to monitor
>>     it.  This will require SNMP be installed or I have done it with simple
>>     scripts or 'C' programs.  Just get the format of an SNMP trap and use
>>     that.
>>
>> 4) You might want to set up the second tty and use that as another monitor
>>     interface.  I would run a simple bash script that would:
>>       'nohup tail -f /var/log/message > /dev/ttyS1&'
>>
>> 5) Add a USB serial port and start a getty on that in your inittab.
>>
>> 6) Attach a USB keyboard or Keypad.  This will allow input only but you
>>     could use it to enter a command for debugging.
>>
>>    Lots of options.  Let me know if I can help.
>>
>> -- 
>> William Estrada
>> MrUmunhum at popdial.com
>> Mt-Umunhum-Wireless.net ( 64.124.13.3 )
>>
>>> Hi,
>>>
>>> 	I have a 4801 I bought about a year + 1/2 ago. It
>>> runs FreeBSD off a CF card with a USB flash to help it. It
>>> ran great for 10 months, but over the last 8 months its
>>> been "locking up" on me. I put that in quotes because I'm
>>> not sure if its the OS or the unit. I do see POWER on
>>> and the NET light blinking. No ERROR or DISK. When I try
>>> to serial console in, I get nothing. 
>>>
>>> 	Is there any way anyone knows to find out if the
>>> OS has gone south, or the unit is not operating properly.
>>> Is there some sort of key sequence to drop into COMBIOS
>>> temporarily or something?
>>>
>>> 			Thanks, Tuc
>>
> 
> 


More information about the Soekris-tech mailing list