Thursday, January 20, 2011
I had some kinda fun today - at work.
We had a fault at the telescope that at first glance was at least odd. Something java (language) would connect to some tasks but not others. The C (language) tasks would interconnect but the java was intermittent at best. Today my favorite sysadmin and I got a chance at trouble shooting. It took forever, but in the end it looks like the solution is that some things are less multithreaded than you'd expect. Making a (that) virtual machine having one (virtual) CPU instead of 2 appears to have solved the problem. I'm glad we did this trouble shooting in the UKIRT remote OPs office. We would never have had the peace and quiet to do this otherwise. (to (the spirit of) David Beattie: no, I didn't have any trouble shooting experience then. I sure do now.)