Re: [Soekris] spontaneous reboot with large packet flows on net5501+lan1741

2015-02-18 Thread Nix
On 18 Feb 2015, Conrad Kostecki told this:

 Please note that there is limited power available for the three PCI expansion 
 connectors.
 There are only 20W available on the 3.3V power pins and 5V pins combined.
 If a 2.5” hard disk is used, it will also need to share the available power.
 An onboard DC-DC converter supplies +12V @ 0.3A and 12V @ 0.1A to the PCI 
 connector.
 If the board is powered by 12V then a bypass circuit will supply up to 1A of 
 the 12V to the PCI connector.

Still... power-wise, the only things I have plugged into mine are the
lan1741 and a Simtec Entropy Key into the USB port.

I'll do a power monitor of the thing with some major network traffic
this weekend and see what happens.

-- 
NULL  (void)
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech


Re: [Soekris] spontaneous reboot with large packet flows on net5501+lan1741

2015-02-18 Thread Nix
On 17 Feb 2015, n...@esperi.org.uk spake thusly:

 On 17 Feb 2015, Andrew Atrens told this:

 It's almost certainly a power issue as power draw for the lan card
 will not be a static thing - ie will increase when transmitting
 packets vs idle.

 Yes, only I'd expect the increase in power draw to be the same when the
 built-in ports fire up, and *those* work fine. Does the LAN card draw
 more power than the built-in ports for a given amount of work, or
 something?

As of a few minuts ago, I can verify that there's enough power to
really, really run the built-in ports hard even when a lan1741 is
installed. I was maxing it out for hours over two built-in ports,
~15,000 packets a second on each port, ~8MiB/s, pretty close to line
speed, plus normal (a few packets/sec) ADSL usage on the other two
built-in ports. The machine took it happily, even though software
interrupt CPU usage often spiked to nearly 100%.

This sort of load would have rebooted it within moments if the lan1741
had been participating.

-- 
NULL  (void)
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech


Re: [Soekris] spontaneous reboot with large packet flows on net5501+lan1741

2015-02-18 Thread Conrad Kostecki
The net5501 manual states:

Please note that there is limited power available for the three PCI expansion 
connectors.
There are only 20W available on the 3.3V power pins and 5V pins combined.
If a 2.5” hard disk is used, it will also need to share the available power.
An onboard DC-DC converter supplies +12V @ 0.3A and 12V @ 0.1A to the PCI 
connector.
If the board is powered by 12V then a bypass circuit will supply up to 1A of 
the 12V to the PCI connector.



So this is much lower, than you could power with 5A @ 12V..

Cheers
Conrad
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech


Re: [Soekris] spontaneous reboot with large packet flows on net5501+lan1741

2015-02-18 Thread Christopher Sean Hilton
On Wed, Feb 18, 2015 at 02:21:22PM +, Conrad Kostecki wrote:
 The net5501 manual states:
 
 Please note that there is limited power available for the three PCI
 expansion connectors.  There are only 20W available on the 3.3V
 power pins and 5V pins combined.  If a 2.5” hard disk is used, it
 will also need to share the available power.  An onboard DC-DC
 converter supplies +12V @ 0.3A and 12V @ 0.1A to the PCI connector.
 If the board is powered by 12V then a bypass circuit will supply up
 to 1A of the 12V to the PCI connector.
 

Thank you! From this I'm can guess that my issue may not be the OS
upgrade from OpenBSD 5.2 to 5.5. It might be that when I upgraded to
5.5 I also turned on the automatic CPU throttling. It might be a good
guess that my power situation was marginal before and that allowing
the CPU to throttle up to higher speed is occasionally dropping the
power below what the level where the PCI bus remains stable? That
will be rather easy to test. One thing I do know is that the Intel
gigabit boards generate a lot of heat. In fact the heat is the reason
I upgraded to the larger rack mount case.

-- 
Chris

  __o  All I was trying to do was get home from work.
_`\,_   -Rosa Parks
___(*)/_(*).___o..___..o...ooO..._
Christopher Sean Hilton[chris/at/vindaloo/dot/com]
___
Soekris-tech mailing list
Soekris-tech@lists.soekris.com
http://lists.soekris.com/mailman/listinfo/soekris-tech