ADMIN: it wouldn't be pyweek...

Without the smell of burning servers.

Yep, this was the last I saw of the server as the deadline approached this morning:

[richard@www ~]$ uptime
 22:09:33 up 29 days, 19:21,  1 user,  load average: 5.92, 5.13, 3.54
[richard@www ~]$ uptime
 22:18:10 up 29 days, 19:30,  1 user,  load average: 11.11, 4.97, 3.85
[richard@www ~]$ uptime
 22:23:00 up 29 days, 19:35,  1 user,  load average: 28.97, 17.25, 9.24
[richard@www ~]$ uptime
 22:25:34 up 29 days, 19:37,  1 user,  load average: 54.62, 31.18, 15.57
[richard@www ~]$ uptime
...
That's me going into a mild panic at the end there. Around 9 hours later I could connect to the machine again.

I've got the logs to try to figure out what happened. At first glance, it looks like the server ran out of memory, and dealt with that problem very, very poorly.

So, sorry about that!

All those who couldn't upload at the last minute should be able to upload now using the files that match the MD5 hashes they sent me. If anyone wasn't on the IRC channel or otherwise didn't think to send me the MD5 hash, please email me (see the site footer).

I'm going to talk to my hosts to see what, if anything, might be done to avoid the problem next time.

(log in to comment)

Comments

Aaand there was a bug in the MD5 uploading ... fixed now. I need sleep.