Cumulus Support

I only noticed the problem because all of my graphs had not moved for two hours, I checked the USB connections by re-inserting the plugs and restarted Cumulus, normal service resumed. I have attached the diags file and you will note from 16:47 "Error!!! Failed to write to USB!".

I received no indication of the problem from either Cumulus or Toolbox and wondered if this is correct and, if so, could Cumulus be enhanced to issue an Alarm.

You may have a badly soldered socket on the console and just re-inserting the cable made the connection for you.

Super-T wrote:You may have a badly soldered socket on the console and just re-inserting the cable made the connection for you.

Hi Terry,

I think it's because I had been moving things around, USB etc (I'm sure I initiated the error).

My concern isn't that the problem occurred but that it was luck that I noticed it when I did. An Alarm Bell from Cumulus (as it was recording a fault) would be great in drawing my attention to the problem as soon as it had registered would be really nice.

There are a number of errors which only get logged to the diags file; some of them do also cause the error light to be flashed. I've been thinking about making more of them flash the error light, but I have no way of knowing if particular errors are commonplace and transitory, and hence crying wolf too often would defeat the object.

I've added 'flash the error LED' for this particular error to the next build.

steve wrote:There are a number of errors which only get logged to the diags file; some of them do also cause the error light to be flashed. I've been thinking about making more of them flash the error light, but I have no way of knowing if particular errors are commonplace and transitory, and hence crying wolf too often would defeat the object.

Yes you're right and I guess a momentary sensor contact loss (in alarm section) or an FTP error (flashing red light) are cases in point - but these do always cause me to register a potential issue. I don’t think this error is common place (only one reference came up in forum search) but when it happens everything still appears to be working OK until you register the fact that the data hasn't changed.

I'm no programmer but a flag set and checked again some minutes later and if it's still set may reduce the crying wolf – but this is a routine easy to suggest but probably far more difficult to implement.

steve wrote:I've added 'flash the error LED' for this particular error to the next build.

I've just noticed your reply while I was responding - You're a star