jmtd → log → geek → hardware → I'll keep this short

kernel: ata1: error=0x84 { DriveStatusError BadCRC }

Lovely. Asking around, it looks like this might be a result of the heat and humidity we're experiencing. Symptoms here include some binaries (various bits and pieces of apt; gzip; firefox) being corrupted (but seemingly not filesystem metadata). The machine remounted the filesystem read-only once this happened.

My brain is suffering a similar fate, but here are a few rants:

1. Why does the evbug module get loaded by default, if it exists? It totally spams the heck out of syslog
2. Why does Debian's syslog, by default, multiplex log messages across several files? My 150MB of evbug log rubbish is occupying 450MB in /var/log.
3. Why does gconfd insist on repeatedly logging the fact writes are failing to a read-only filesystem? It makes it particularly difficult to find the cause when you have to filter out this first.

One thing that has just occured to me; how was gconf logging anyway, if the filesystem is read-only?

I still have the sneaking suspicion this could be software, not hardware. I haven't experienced any such problems in Windows on the same machine yet. Still, I'm mentally calculating how many DVD-Rs I will need for a panic backup (40, it seems).