Copying dumps from Postgres to MonetDB

When moving bulk data from postgres to monetdb you can use the psql \copy directive and the mclient copy like so:

psql:
\copy gftp_packets to 'my dump file' with csv
mclient:
copy into gram_packets from 'my dump file' delimiters ',','\n' ;

Booleans don't work, but a little python will fix it up:

fin = open('my dump file', 'r')
fout = open('my new dump file', 'w')
falserep = (x.replace(',f,',',0,') for x in fin)
truerep = (x.replace(',t,',',1,') for x in falserep)
for line in truerep: fout.write(line)

I was too lazy to fix the problem with successive booleans, so I just rinsed and repeated until all the t's and f's were 0's and 1's.

Google Fiber for Communities

If you are a Bloomington area person please let me know. I am trying to organize a bid for the Google Fiber for Communities experiment. http://www.google.com/appserve/fiberrfi/

Bonnie++ and Xen

I ran some filesystem tests with bonnie++ on a DomN, Dom0, and the machine that is exporting the lun over iscsi. The results can be found here. Looks like there is some significant impact from using Xen on older cpu's without the extra virtualization goodies. I suspected this would be the case but now I have some hard numbers.

14:59 Remaining

Somebody noticed a comment I made and turned it into a whole blog entry. Looks like I've got 14:59 remaining.

Rockafeller/Gerhartsreiter Half Life 2 ?

Anybody notice that Rockafeller/Gerhardsreiter looks a lot like the guy on the cover of the Half Life 2 box?

Considering EC2 interface

I am considering implementing a knock off of the EC2 query interface and dropping the whole ajaxy web interface. I can probably do this faster and better than trying to do everything in a browser. I only need to work out the aws id and aws secret key analogs and display those in the web interface. My plans are to still use myproxy usernames and passwords for authentication. Still don't have a good way to communicate with the hardware nodes, but I am thinking about xmpp or some other message queue. It doesn't look like anybody has written any documentation for using libvzctl which would have been nice, I don't like the idea of just running the commandline scripts and capturing the output it seems like too much of a perl hack.