Hello,
I would like to inform you about a problem encountered upon writing in a sqlite database on a multi-core machine (Sun T2000). Indeed, on T2000 systems, during endurance tests that induce a lot of regular writings during several hours in a sqlite database (3.3 but also 3.6 sqlite version), the following error is returned by sqlite; "SQL logic error or missing database". This error happened first cyclically (every about 2 hours or 3 hours, depending on endurance tests) with -DSQLITE_THREADSAFE=1 as thread-related sqlite compilation option. Now, with same -DSQLITE_THREADSAFE=1 and new DSQLITE_ALLOW_XTHREAD_CONNECT=1 compilation options, the error happens only 1 time during the test. A new test with -DSQLITE_THREAD_OVERRIDE_LOCK=1 option instead of -1 is planed. To note that this error doesn't occur with a mono-core system (Sun Netra 240), and that the program that writes in the database isn't multithreaded. If you have explanations or/and some correction ideas about such an error that happens in T2000 machines (multi-core) and not in Netra 240 machines (mono-core), feel free to share them with me. Thank you, José Nogueira _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
