Correct. In rollback journal mode when one connection says "I'm ready to write now" it blocks any new transactions from being made, but it can't do anything about existing read transactions. It has to wait for them to finish their reads and end their transactions before it can do any actual writing.
In WAL mode, when one connection says "I want to write now" then as long as no other connection has called dibbs on writing, it can start writing without worrying about existing readers. -----Original Message----- From: sqlite-users [mailto:sqlite-users-boun...@mailinglists.sqlite.org] On Behalf Of Gerlando Falauto Sent: Tuesday, August 07, 2018 8:54 AM To: SQLite mailing list Subject: Re: [sqlite] Database locks I just realized I'm using default settings... perhaps I should use WAL mode instead? Thanks, Gerlando On Tue, Aug 7, 2018 at 2:22 PM, Gerlando Falauto <gerlando.fala...@gmail.com > wrote: > Hi Simon, > than you for your answer. > > On Tue, Aug 7, 2018 at 2:09 PM, Simon Slavin <slav...@bigfraud.org> wrote: > >> On 7 Aug 2018, at 12:55pm, Gerlando Falauto <gerlando.fala...@gmail.com> >> wrote: >> >> > I'm trying to implement a logging system based on SQLite, using python3 >> > package apsw. >> > There's one process constantly writing and another one reading. >> > From time to time I get an exception from the writer, complaining the >> > database is locked. >> >> Please set a time of at least 10,000 milliseconds for /all/ connections, >> both reading and writing: >> >> Connection.setbusytimeout(10000) >> >> <https://rogerbinns.github.io/apsw/connection.html?highlight >> =timeout#apsw.Connection.setbusytimeout> >> > > Hmm... are you saying the writer could potentially block for up to 10 > seconds? > If that's the case then I should rethink the whole logging process cause > it might end up losing incoming data if waiting for too long. > In any case, I still don't understand whether the reader would block the > writer or not, and in what phase. > A reader could potentially take a long time (even longer than 10 seconds) > to read all the data... > > >> If you're already doing this, please post again, telling us whether >> you're using two separate connections or passing the connection handle from >> process to process. >> > > It's two separate connections. Is that bad or good? > > Thank you, > Gerlando > > _______________________________________________ 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
