[atlas] Probe offline

On 07/03/2016 10:49, Gert Doering wrote:
> Hi,
>
> On Mon, Mar 07, 2016 at 10:39:47AM +0200, Hank Nussbacher wrote:
>> What does that mean?  I can try reseating the USB again, but if that
>> doesn't work, it could be the USB is fried?
> Try the USB stick in a "normal" PC and see whether it can be formatted
> there.  I recently had one of mine completely break - the stick could be
> seen, but it was empty and all write access failed.

I pulled the USB stick and tried formatting it.  Even though it says 4GB
Sandisk, I could only get it to 1GB.
So I opened a new probe, extracted its USB stick and stuck it into the
probe as well (unformatted).   Still off-line.

I went to our "lights out" facility 3x today - a 15 minute brisk walk
across campus and don't have time to
go there again. 

At home it is far easier to play with these things then it is when the
probe is installed as close to your network core as possible (which is
usually at a LO facility).  I know exactly how you feel!

-Hank

>
> I'm not sure what the Atlas v3 does with its USB stick, but this is the
> number one problem issue...  maybe a new firmware version could be designed
> that has more advanced flash handling (like, ubifs instead of "normal"
> filesystems) and falls back to "not use flash if the flash is broken".
>
> What I see with my probes is that the aim of the flash buffer ("we can 
> store measurement results if we can't upload them to the control server
> due to network outages etc." -> less probability of result loss) is 
> actually backfiring into "extended downtimes of probes due to USB breakage
> of probes in locations where you can't just-so swap the USB flash"...
> (two of my 3 v3 probes have had virtually no network outages since they
> are operating, and the central servers also had few outages - but both
> have been down for weeks because I just had no time to go out, buy a 
> new flash drive, and *drive over* to replace it - once again)
>
> Gert Doering
>         -- NetMaster