Sunday, June 01, 2008

Incident ID: 4042264

"Due to its complex nature, your issue has been relayed to our Advanced Technical Support Team. Our most skilled technicians will be working to resolve your issue quickly and completely. You will be notified promptly upon resolution."

Great. Just... great.

Like I mentioned last time, the map likes to make a liar out of me whenever I say it's working great. And it is working great. When it works. And it works all the time. But now it seems I have run into Someone Else's Problem.

It turns out that "Someone Else" is probably the Usual Suspect.

Randomly last Sunday, I would look at the map and it would be blank. Not whitespace blank, not 404 blank, but markerless. That should never happen. Even when no one is on the server, the last uploaded data file should still be there and at least one marker should be displayed until another player shows up.

Here's what is supposed to happen: once every 30 seconds or so a bash script wakes up, checks who's playing, their score, ping and which server they're playing on through the UT Webadmin interface. Then it looks up the latitude and longitude of the player's IP address, and sends all this information in a tiny, tidy XML file, via FTP, to the Web server at where it is displayed through the map using the Google Maps API.

It turns out the weakest link in that strategy is the FTP protocol itself. Or rather, the FTP server at the other end.

As I dug deeper into this issue I found that the XML file was getting stomped on by the server. It was a zero byte sized file every time the map was markerless. So I ran tcpdump to watch the file transfer. It was working, the file was being uploaded and it wasn't a zero byte file (in fact there is logic in the script that won't even send a zero byte file).

The server was responding like this:

426 Connection closed; Transfer aborted.

In non-technical terms this means "FUCK YOU I DON'T WANT YOUR GODDAMNED FILE".

That was last Sunday. I hacked things around on my end and nothing helped. I monitored this all week and discovered it seems to happen at about 20 minutes after the hour and persists for five to thirty minutes. Then it goes back to normal.

Last night I opened Incident ID: 4042264 with GoDaddy. But I think I've already determined the root cause, and it will come as no surprise to anyone.


Fuck. We're doomed.


Here is GoDaddy's somewhat worthless response:

Dear Sir/Madam,

Thank you for contacting Hosting Support. We have resolved the issue with the FTP on your hosting account.

We were able to connect both internally and externally from our network to the hosting account. You should now be able to connect to your account. We apologize for any confusion regarding this issue.

Please contact us if you have any further issues.

Ben A.
Hosting SupportHosting Operations

This is fairly typical tech support horseshit. "We can do it so it's fixed!" They completely misunderstood the issue, probably on purpose. Ignore it and it will go away.

I wonder who they think is confused?

No comments:

Post a Comment