On Thu, Nov 7, 2013 at 7:57 PM, Simon Slavin <[email protected]> wrote:
> > > I will definitely look into the wal mode. Are there any test results for > disk failure robustness of this mode compared to journal mode. > > This won't help. As long as you're using a network file system which does > not support fsync() properly you will get more corruption. > It does help, because the there are fewer times in WAL mode processing where the order of write operations matter. Hence, there are fewer opportunities for a power failure or crash to cause problems. But it is not perfect, and corruption can still occur. The only way to reliably prevent corruption is to use a filesystem that never reorders writes across an fsync() call. The previous paragraph assumes that the cause of your corruption is a bug in the filesystem that causes writes that happen before fsync() to actually occur after one or more writes that were issued after the fsync(). If the cause of corruption is something else, then WAL mode might not help as much. -- D. Richard Hipp [email protected] _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
