I'm using sqlite from the C API on top of EC2 where EBS will sometimes
give me IO errors. For the most part, this is OK and my application can deal
with them, but I'm not entirely sure *how* to deal with them correctly in all
cases.
My usage is pretty simple:
begin transaction;
-- do a bunch of inserts (on conflict replace, so many are
updates in practice)
commit;
Whenever an insert fails, I can requeue it to happen at a later time.
Right now, when a commit fails, I just keep retrying it until it succeeds.
The types of problems I might have that could lead me to failing to
commit should theoretically also lead me to be unable to rollback. My app must
assume that these errors in the IO layer will eventually heal.
What's the best course of action to keep my application running and try
my best to keep my on-disk data sane?
(a variation of this was also posted to SO and may have some useful commentary
around it:
<http://stackoverflow.com/questions/2842752/how-should-i-deal-with-sqlite-errors>
)
--
Dustin Sallings
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
