El 13/10/13 16:57, Bastian Bittorf escribió:
>>> the board is ok, no need for replacing it. we have several 1000
>>> boards,
>>> and we see that everyday some of them are hanging in this situation
>>> (right after nightly reboot). statistically that means:
>>>
>>> if you reboot a board ever night, than it will 'hang' once a year.
>>> also read http://en.wikipedia.org/wiki/Soft_error
>>> and http://en.wikipedia.org/wiki/Single-event_upset
>>>
>> Hm. In this case, it matters. From your first message, I realized that
>> the same board constantly can not boot. And in such situation, would
>> be nice add capability of reboot to the mainstream loader.
>>
>> Allow me to ask, why do you reboot the equipment every night?
> 
> the only reason is 'cleanup' and fallback to normal operation if
> something very stupid happens. we deciced to use this "russian method",
> after realizing, that there is always one error more we did'nt catch.
> 
> we use a 'maintenance window' at ~4 o'clock to do that. it has
> more advantages, than disadvantages...


we came to the same conclusion and reboot every day at 5AM. There are
only 40 nodes but we never saw the symptom you described yet; maybe it's
just statistic luck :P
_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Reply via email to