A few more tests were tried on the offending hard disk, with mixed results. Then the disk was re-connected and booted up. The root partition came up clean, but the /usr partition would not, failing fsck repeatedly even after reporting "filesystem is clean".
Finally I gave up and ran diskroaster, which reported five errors. Then I tried dd, sending output to /dev/null. That reported success. Eventually I stumbled on the correct incantation of smartctl, which didn't report any obvious faults. In particular "current pending sectors", which I understand to be indicative of problems, was zero. The results of diskroaster, dd and smartctl are in http://www.zefox.net/~fbsd/rpi2/disktrouble/ if anybody's willing to have a look. At this point dd and smartctl seem to see nothing wrong, diskroaster reports five errors but the last has an extra byte in the block offset, which seems odd. Also, the runtime values given were wildly off, two hours turned into all day and into the night. At this point smartctl -t long /dev/da1 is running. Any guidance in what to look for would be much appreciated. It'll finish in about three hours' from this writing. Thanks for reading, bob prohaska
