What a week.
I think that the large cluster may live, though it’s still not good. We’re actually writing a butt-covering letter that says “we tuned and configured your cluster. It’s still not a good architecture. We think that it will crash frequently, costing you lots of time and frustration. We’ve given you suggestions for improvements that you could make to change this, and you have decided against implementing those improvements. Therefore, we wash our hands of this nightmare, and walk away.” Yeah, it’s that bad. Beyond having a poor setup (a 240 node cluster whose network is a series of 48 port 10/100 NetGear switches daisy-chained through the uplink ports) they have insufficient cooling and power. They’ve already popped the main electrical into their building, twice. The temperature in the server room regularly exceeds 100 degrees.
Had a good meeting yesterday with my whole company, in which we discussed the fact that we’re running flat out with current commitments, and we’re about to open up a bunch of new markets. This is insanity. Summary: Communication is good.
The plan for the weekend is to clean, and then to possibly go see the Halloween display at the zoo, or cruise into Boston to see the Lord of the Rings exhibit at the Museum of Science.
Leave a Reply