[Dx4win] LoTW Upload - Filenames

[Rick Murphy]

On Mon, Dec 10, 2012 at 5:37 PM, R.T.Liddy <k8bl at ameritech.net> wrote:
>
> It's a pretty common way of doing it and I'm sure a lot of folks probably
> used a similar system. But, then I read that one of the bugs at the
> ARRL LOTW computer was that if a Log Submission came in with the
> same name as an earlier one that hadn't been processed yet, the new
> one would replace the earlier one in queue. So, if my L120112 came in
> before John Doe's L120112 file, mine would be deleted from being entered
> into LOTW and I wouldn't know it unless I was closely tracking everything.
> I might never know if a lot of Q's were missing.
>
> Now, I use a file-naming convention that would be highly unlikely used by
> anyone else. It's possible, but highly unlikely.

You're misunderstanding the LoTW bug. It doesn't matter what you
called your files.

The LoTW developers realized that people could have duplicate file
names, so they devised a scheme that renamed user files to a form
"msg-xxxxx-x.tq8" where the "x" characters were numbers. The number
was derived from the UNIX process ID, leaving a range of 65,535
different values for the file name. As long as the queue of files
waiting processing didn't get very long, all was well. However, once
things started to take multiple days to process (even more than a
couple of hours, actually) the chance of a collision goes up. There
wasn't anything you could do to avoid the bug, nor anything you could
do to make it worse (other than perhaps submitting a log when there
was a big backlog,)

I've had at least one log lost due to this bug back before it was
discovered. Luckily, I keep track of before/after counts so I'm pretty
sure nothing else was lost.
73,
    -Rick
-- 
Rick Murphy, CISSP-ISSAP, K1MU/4, Annandale VA USA