Transaction delays apply to read as well. SQLite places a lock while reading too, to ensure the database is intact during read. Otherwise tables will half half complete rows.
Read "begin transaction", difference between immediate and exclusive transactions. Roman Sent from my T-Mobile 4G LTE Device -------- Original message -------- From: Peng Yu <pengyu...@gmail.com> Date: 10/16/19 6:41 AM (GMT-05:00) To: SQLite mailing list <sqlite-users@mailinglists.sqlite.org> Subject: Re: [sqlite] Network file system that support sqlite3 well > I know for sure that IBM's GPFS guarantees locking. I think GPFS is "global > parallel file system". It is a distributed file system. But it will be > rather slow. If only few jobs run in parallel, all will be ok. Locking will > always guarantee database integrity. > > With lots of jobs, you will see you have to increase sqlite timeouts to > hours. Waiting for a lock will be much longer than transaction, obviously. > If transaction takes 1 second, with 1000 jobs, you will need timeout of > 1000 seconds. But locking adds very large overhead. Timeout will have to be > 2 or 3 times that. > > If GPFS is loaded by other jobs (from other users, not even sqlite users) > the wait times will increase. Although I have many compute nodes, there is most only one computer node write to the DB file. There can be several processes to read the same DB files from each compute node. I guess the transaction delay only applies to write? Does it apply to read? -- Regards, Peng _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org https://nam02.safelinks.protection.outlook.com/?url=http%3A%2F%2Fmailinglists.sqlite.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Fsqlite-users&data=02%7C01%7Croman.fleysher%40einstein.yu.edu%7C1c9c57629ef043b5847d08d7522562fc%7C04c70eb48f2648079934e02e89266ad0%7C1%7C1%7C637068192835821781&sdata=bqetTp5ZAa4ZJEDJiv6MDIyWZ2IahgUHQ5gATgfg2u0%3D&reserved=0 _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users