[EMAIL PROTECTED] wrote: > > Process A wants to modify the database, so it flock()s > the rows it needs to changes and starts changing them. > But half way in the middle of the change, somebody sends > process A a SIGKILL and it dies. The OS automatically > releases the flocks as process A dies, leaving the > database half-way updated and in an inconsistent state, > with no locks.
When Process A is being killed and OS automatically releases the flocks, will rollback journal be present or not ? I think, in this situation, hot journal will be left when Process A is killed. > Process B comes along and opens the database, see the > inconsistent state, and reports database corruption. > If Process B finds a rollback journal...then...it can use that to rollback database and bring it to consistent state. ------- Curiously thinking on this... What if a new SYSTEM-LOCK-MGMT-TABLE is added in SQLite database managed by itself for locking management ? (Surely, it is not an easy task to manage SYSTEM-LOCK-MGMT-TABLE..!!! ) Assuming that, if there is one an added SYSTEM-LOCK-TABLE in SQLite database managed by itself for locking management. Then when Process A wants to modify the database and it flock()s the rows it needs to changes, it should add corresponding info as record in SYSTEM-LOCK-TABLE for that... When somebody sends process A a SIGKILL and it dies, OS will automatically release flocks as process A dies. Process B (any other process accessing database) can find necessary locking related info from SYSTEM-LOCK-TABLE, as well as hot journal and so Process B can bring database back to consistent state. Rohit -- View this message in context: http://www.nabble.com/SQLite-Vs-VistaDB---Comparison-----tf1797052.html#a5255158 Sent from the SQLite forum at Nabble.com.
