Mike Wilson's Blog


I was a MA/MST student at Stanford's Center for Computer Research in Music and Accoustics.

This is my weblog.

Click here for my main page.



2012-05-07

I had a bit of a scare today with my primary laptop.  I had removed
the hard disk to see if I could fit a different, physically larger
drive in the bay.  It wouldn't fit, so I put the original drive back
in.  When I tried to boot I got an error about grub not being able to
locate a menuentry, and none of my kernel images would boot.  Whoops.
Did I physically damage the drive somehow?

Fearing the worst, I booted with a live USB stick and ran fsck on all
the drives.  No errors.  I diligently connected an external drive and
backed up all of my data before doing anything else.  Everything
seemed to be going smoothly.

Once all the data was backed up I started to poke around a bit more.
I have a Windows partition, and I was able to boot to Windows.
Interesting.  I then remembered that I had installed a new kernel just
before shutting down and removing the hard drive.  And I had been
messing around with the grub configuration file (/boot/grub2/grub.cfg)
because I had recently installed a memtest target and the script to
update the config file also dumped a bunch of unnecessary things in
there.

Well, to cut an already too long story short the new kernel didn't
format its menuentry correctly, probably due to my editing the config
file and it not being able to pattern match on something it was
expecting.  I was able to get to it with the live USB stick and fix
the formatting, and everything booted up fine (and running another
fsck didn't show any issues).  Looks like the hardware is fine!

How does this relate to computer music?  Well, my synthesizer and all
of my work is on this machine.  I maintain regular backups, but
imagine if I had to play a gig and something like this happened.
Wouldn't be pleasant.

There's a trade-off between flexibility and stability.  A
general-purpose system such as a full-blown laptop computer is very
powerful, but very complex.  It can fail in weird and interesting
ways, and no matter how reliable it is there are so many variables
that it's virtually impossible to predict them all.

A specialized hardware system like a rack-mount tone generator will
generally be more reliable (or at least predictable) but it's hard to
do active development on something like that.

I heard from some of my classmates that some professional touring
groups that use computers will have multiple redundant systems, and
hire someone specifically to smoothly fade between them if one of them
dies during a performance.  It seems like something like that's
probably the way to go if failure is not an option.

As a student I don't quite have the money or need to do something like
that, but it's always good to keep in mind that things can go terribly
wrong at any time.  I made sure not to update my system at all in the
week before Modulations and I still had some unexpected behavior when
I connected to the projector right before my performance (every other
time I've done that it's come up in multiple display mode, but of
course when it counts it came up in mirror displays mode).

Anyway, back up your data!


email mwilson@alumni.caltech.edu
Disclaimer: the views herein are my own and do not represent the views of Stanford University. All material copyright Michael J. Wilson.