On Sat, May 18, 2013 at 4:41 AM, thanumalayan mad <[email protected]>wrote:
> > Expected result: You always find that the transaction had been executed. > Observed result: You sometimes find that the transaction did not execute. > The core team has discussed this. In order to avoid a substantial performance hit against transaction COMMIT, we have chosen to not do fsyncs on the directory when a file is unlinked, and thus to allow loss of durability following a power loss event. ACI without the D is still guaranteed. But not the D. The overwhelming majority of applications care not one wit about durability following power loss. For most applications, it is sufficient that the file is uncorrupted. If recovery gives you a snapshot of the file as it existed 5 seconds prior to the power loss, that's fine. WAL-mode transactions should be durable across power-loss events. So if durability is vitally important to you, you can always set PRAGMA journal_mode=WAL. Are you observing loss of durability following power loss in WAL mode? Is there any place in the documentation that we have overlooked where SQLite claims to be durable across a power loss in rollback mode? -- D. Richard Hipp [email protected] _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
