View Single Post
  #8   (View Single Post)  
Old 14th December 2010
rocket357's Avatar
rocket357 rocket357 is offline
Real Name: Jonathon
Wannabe OpenBSD porter
 
Join Date: Jun 2010
Location: 127.0.0.1
Posts: 429
Default

Quote:
I made as I could, since it works, where is the problem...? ;-)
I had a job interview at Google last summer, and they asked me the question "You have a command to run on 100,000 machines. How do you approach it?" I wrote a very classy threaded Python program to utilize their existing key-based infrastructure to ssh simultaneously to all 100,000 machines as fast as my workstation could launch child threads. As each child thread successfully established an ssh connection, it would issue the command to the machine it was connected to. The program would then gather the results from each child thread in a directory hierarchy on my local machine so I could grep -R the output for issues. It was awesome....but it answered the wrong question. They asked about my approach (which by my program I implied "I'd just run the command, f*ck it"). They were looking for "I'd run it on a single test machine and wait a bit to see if it caused problems. If it appeared to be ok, I'd run it on a bigger group and wait again. If that was ok, I'd write a program (much like my python code) to run it on everyone."

Getting compliments from Google engineers about my coding style was nice, but I didn't get the job. Point is, it takes more than "it works" for it to be correct.

Last edited by rocket357; 14th December 2010 at 05:08 PM.
Reply With Quote