I think it does at some point. I’m at home right now so I have to check this again tomorrow when I have access to the source.
Should there be rbu calls if the application is only _reading_ fro the database and not updating? > Am 19.09.2018 um 21:48 schrieb Dan Kennedy <danielk1...@gmail.com>: > > On 09/20/2018 01:49 AM, Roger Cuypers wrote: >> I'm trying to optimize a C++ application that uses sqlite 3 for database >> access. As far as I know it uses journaling with WAL and has a lot of >> files/tables (about 400). Profiling this application with Linux perf, I >> found that it spends about 30% of its time inside the rbuFindMaindb function >> of SQlite3. This function mostly consists of a loop that goes through all of >> the journaling WAL files in the virtual file system, so it seems that in >> order to bring down the cost of said loop I would have to reduce the number >> of WAL files. Seeing that apparently sqlite creates a WAL for every database >> file, I'm not aware how to fix this. >> >> Now my question: How can I optimize this? Can I reduce the number of WAL >> files without reducing the number of database files? Is there a different >> journaling mode that does not call rbuFindMaindb so often? Can I optimize my >> program so that this function is called less often? > > Are you actually using the RBU extension? > > https://www.sqlite.org/rbu.html > > Has your application called any sqlite3rbu_*() APIs? > > Dan. > > _______________________________________________ > sqlite-users mailing list > sqlite-users@mailinglists.sqlite.org > http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
