Hello, I'm running a web service that uses SQLite that throws a disk I/
O exception every once in a while, meaning once every few weeks.
Details: SQLite 3.7.0.1, being called from an Apache/Django/Python
multi-threaded application running on Solaris 10. The database file
is on a local filesystem, and is <200MB. Disk space is not an issue
on that filesystem.
Everything appears fine, and then one transaction out of the blue gets
a disk I/O error when doing a "BEGIN IMMEDIATE". The database passes
an integrity check. These disk I/O errors don't seem to correlate
with anything. When one happens, there isn't a lot of activity on the
system (individual transactions occurring once every few seconds, say,
so threading/locking/contention wouldn't seem to be an issue), and the
transactions immediately before and after the failed transaction
succeeded just fine. If there is any correlation, they all seem to
happen in the middle of the night. I've checked with my sysadmin, and
he can't think of anything (backups, virus scans, etc.) that might be
happening at that time.
I've searched for help on this topic and have come up with nothing.
It should be clear from the above that directory permissions are not
the problem.
Any ideas? Unfortunately, the (standard) Python SQLite wrapper I'm
using doesn't provide access to any more information (if there is any
to be had).
Thanks in advance,
-Greg
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
