<![CDATA[Mahalo's real talent hunt - Valleywag Comments]]> <![CDATA[Mahalo's real talent hunt - Valleywag Comments]]> <![CDATA[Mahalo's real talent hunt]]> @ClosingTag: You're hired!

vulturesquadron

]]> <![CDATA[Mahalo's real talent hunt]]> Talent show winner(s): Taurus and Fondue.

No question. And BTW, I bet they can do engineering, too. They're British. Probably trained at Oxford with Jonathan Zittrain.

Rachel Marsden

]]> <![CDATA[Mahalo's real talent hunt]]> Oh god, that is hilarious. 8 million pageviews and he wants someone with Hadoop experience? You could run that site off one single-core server if you needed.

BTW, here are the answers to Calacanis's questions:

1. What is wrong with this picture? YOU'RE USING NFS. Nevermind the obvious permissions problem (Apache running as nobody, and Tomcat running as root), the fact that you're relying on NFS rather than local storage and synchronization is braindead.

2. Expect; although the fact that you're using Expect for something like _rsync_ is even MORE braindead! Set up a trusted key relationship between the two hosts and you never need to type in a password with rsync ever again.

3. Because your database is IO bound with all the binary log writing, dumbass. Quick and dirty solution: innodb_flush_log_at_trx_commit=0 Correct solution: RAID5 battery backup with a nice big ram cache.

4. "Oh shit, we ran out of rows in our table!" If you want a hardware solution, add a few Xeons, but if you're in such trouble that switching to 64-bit architecture actually helps, your schema sucks.

ClosingTag

]]>