Re: SD card corruption

On 7/13/2023 15:12, mike tancsa wrote:
> On 7/13/2023 3:02 PM, Bob Bishop wrote:
>>
>> What’s the environment like? SD cards really don’t like being run hot 
>> for instance. Are there radios or other electrical noise nearby?
>
> Its a messy retail environment, but not too hot. The design of the 
> APUs have excellent passive cooling and I monitor the CPU temp at a 
> few hundred sites and have a good baseline. There is very little 
> variation and failures dont seem to correlate with the few hot 
> outliers. CPU is usually around 50C. 
> (https://www.pcengines.ch/apucool.htm). The recent fail site the 
> weekly avg was 49.9C with almost no variation / spikes
>
> Could exposure to a burst of intense em scramble the SD card ?  I 
> would think other devices would be impacted if that were the case. 
> Whats odd is that I am testing one of the returned cards right now. I 
> wiped it, filled it with 15GB of a few random files and am 
> continuously checksuming the files and they are fine. I would think 
> that if the SD card failed, it would continue to fail. Hence, I am 
> just trying to better understand what causes this. The SD card 
> hardware layer is all "black box" to me
>
I have a decent number of these (and this type of config, on pcEngines 
machines, in the field) and have yet to see this specific sort of 
failure.   Typically when I see SDs fail they go write-locked AND 
corrupted -- and are unrecoverable.

This sounds like a bug in the on-SD-card firmware that is scrambling the 
allocation table.  That's nasty and its also likely out of your 
control.  The only thing I can come up with that might contribute to it 
would be unstable power.

-- 
Karl Denninger
karl@denninger.net
/The Market Ticker/
/[S/MIME encrypted email preferred]/