I was off in the Vancouver area over the last two weeks visiting my fiancée's family, and I got out of the internet/blogging-zone for a while. Now it's time to catch up some ;).

Vancouver was wonderful, as always ;). Flaky wind gave the advantage of maneuverability to the canoe navy, but the sunfish's mass held it in good stead. Despite much sniping and tactical maneuvering, we all ended up in the ocean anyway… several times… that's either a win or a draw, but either way it was a good time :).

Over the weekend I bought a pair of Logitech Quickcam for Notebooks (046d:08dd) so I could video-skype with my fiancée while we're apart for a bit. Like a bunch of other people, I didn't realize that this version of “Logitech Quickcam for Notebooks” was not particularly well supported on Linux, since the earlier version of “Logitech Quickcam for Notebooks” is. Since then I've been running around on my not-to-easy-to-get-source-for EeePC, wring to compile and install Michel Xhaard's gspca drivers.

<rant>

Which is not difficult once you've removed the unionfs junk the EeePC comes with, cleared out a bunch of space, installed the necessary build tools, and generally manhandled your EeePC into something resembling a sane distributions basic installation. Oh, and you'll need a nice external drive, since it's hard to compile a 400 MB kernel in the 150 MB of headroom I had left on my 2 GB Surf after I'd installed those, and emacs, and a few other fun toys :p.

And after I had installed the drivers and had xawtv working fine with the camera, the problems weren't over, since skype pretty much pegs my CPU, which often causes the video to lock… Sigh…

</rant>

Upon my return to work, Marisa expressed interest in the GROMACS Molecular Dynamics simulation software, so we installed in on our cluster and spent Monday learning the ropes. Things seem to be going smoothly so far, although we're not sure if our nominally MPI-enabled mdrun is actually using more than one node, since I haven't found anyone using GROMACS over PBS via our mpiexec (see (Kristina Wanous' note on debianclusters). We'll run some scaling benchmarks soon to convince ourselves that GROMACS is indeed using MPI.

We also ran into some difficulty running the default Debian-packaged RasMol through a SSH tunneled X server. No problem, says I, we'll use Lionel Cons' mxconns, which I had run across earlier while trying to set up a cluster-wide X port for running MPE. However, rasmol somehow manages to lock up so hard that it freezes the whole X server, right through mxconns and the SSH tunnel! I didn't even think that was possible. The current workaround is to run rasmol on your own machine ;).

While discussing the cluster, Marisa pointed out the usefulness of iptables or some sort of firewall on the cluster. I had just assumed it was already behind a conservative router. Stupid assumption. I spent most of Tuesday jumping between the Rasmol X-lock and figuring out how iptables works, and locking out my ssh connection a few times. After an embarassing number of trips down to the cluster room to clear my bad rules, I believe I have things straightened out now. Mostly following the Linux Administration Handbook.

Today I finally got back to work on testing my simulation, and broke down and installed org-mode for emacs. I think I'm in love ;). See, for example, John Wiegley's 's blog on using org-mode as a day planner.