Key collisions from backerkit

[CubePolyhedron]

Got the email, found the key, but looks like someone beat me to it, somehow - don't like the implications of that, but maybe it was an issue with duplicate keys being given out.

[x4000]

Hoookay... I have made a terrible error and uploaded duplicate codes.  Nothing like a little extra incompetence to add some flavor to a mistake I'd already made in under-provisioning, eh?

- This affects 91 keys at present out of the 1091 distributed, and 257 keys that have not yet been distributed to anyone.

- I've emailed backerkit and let them know about this, and asked them to simply remove those keys from their system and take them away from the people's accounts from row 1001 on upwards.  They are on PST, so I imagine it will be a couple of hours, frustratingly -- again, this is my fault, not their, though.

- Once the duplicate keys are out, I can then refill the above-1000 set with fresh ones and re-issue those to the affected people.

- Unfortunately that STILL won't be a guarantee of no problems, because with the 91 keys that are duplicated between people, some of those were registered by people who got sent them on the 28th (batch 2), others were registered by people who got sent them in on the 27th (batch 1).  I have no way of checking which person activated it.

- So there's going to be a further cleanup step that has to happen with some subset of those keys that have been registered, to make sure that someone from batch 1 isn't denied a key because someone from batch 2 registered it first.  I haven't figured out the solution to this just yet, but it's what I'm working on now.  It shouldn't be too bad, but it is one bit I have to sort through.

REALLY sorry about this mess, I can't stress that enough.  It will be resolved today, within the next couple of hours.

[jerith]

Hoookay... I have made a terrible error and uploaded duplicate codes.  Nothing like a little extra incompetence to add some flavor to a mistake I'd already made in under-provisioning, eh?
[snip explanation of muckup]
REALLY sorry about this mess, I can't stress that enough.  It will be resolved today, within the next couple of hours.

And this is what good customer service is about. The inevitable mistakes that happen because humans are human, followed by a quick response and corrective action.



(This kind of thing is also the reason my team has a "no deploys on a Friday" rule. The end of a long week when you're tired is seldom the perfect time to rush something out the door before you go home for the weekend. But sometimes you need to do it anyway.)

[x4000]

I try not to do huge deploys late on Fridays, too.

Right now the data is in the hands of backerkit for the last 2 hours, and I'm just waiting for them to scrub it out...

[x4000]

The data is MOSTLY fixed.  There were previously 182 duplicate keys that needed to be stripped out so that they could be reassigned, and then a further 300-something duplicates that had not been assigned that needed to be negated so they would not be.

All that got taken care of, and people have had new keys reassigned... with the exception of 14 last stinking duplicates that didn't get cleared out in the first pass.  They were in the dataset I sent to backerkit, so it was just something that got munged up in their database calls.  I imagine due to some folks with multiple duplicate keys, if I had to guess.

At any rate, those last 14 are something that are being sorted out as soon as they can (I just emailed them, after re-testing all the data).  If you just got another email tonight with a fresh key, that one is good.  The 14 are remainders from the other data from the 27th and 28th that did not get fully cleared.

For some of you who already registered a key, you'll now find you have an extra thanks to the reallocation of these 182 keys that I duplicated around.  Lucky you!   Consider it my apology for the foul-up in the first place, and no worries about it from this end.

Thanks again for your patience,
Chris

[x4000]

Okay, the last 14 are all straightened out -- that should be everybody all set!

Thanks again for your patience.