After another 20 attempts, I managed to extract (twice) the original error that is triggering the rollback and its failure.
Instead of the SQL logic error or missing database error at a rollback attempt, I now know that the more relevant error from SQLite is some kind of disk I/O error occurred Somehow, this does not feel like progress. The error seems to happen only with read-only database connections, where using read-only connections is the recent change. If anyone with SQLite experience has an idea what might be wrong, I'd welcome advice. (The journal mode is DELETE.) At Fri, 20 Dec 2013 10:44:45 -0700, Matthew Flatt wrote: > Thanks! I've seen this error twice before today, and I've managed to > provoke it twice this morning --- but that's out of about 20 attempts > this morning, so it's slow going. > > At Fri, 20 Dec 2013 09:51:23 -0500, Sam Tobin-Hochstadt wrote: > > There's been an intermittent error on a few DrDr runs on `master`. You > > can see it here, search for 'rollback'. > > > > http://drdr.racket-lang.org/27963/pkg-src/build/make > > > > This doesn't seem to be causing anything else to break the DrDr > > notices, but since there were some DB problems with the release, I > > thought I'd bring it up. > > > > Sam > > _________________________ > > Racket Developers list: > > http://lists.racket-lang.org/dev > _________________________ > Racket Developers list: > http://lists.racket-lang.org/dev _________________________ Racket Developers list: http://lists.racket-lang.org/dev
