On Wed, Apr 26, 2017 at 5:58 PM, Dominique Devienne <[email protected]> wrote: > On Wed, Apr 26, 2017 at 5:34 PM, Kim Gräsman <[email protected]> wrote: > >> Great, that means the numbers add up. This is a monster transaction >> updating 5M rows, and page size is 512 bytes, so I think we have roughly 1 >> row/page. > > Which such a small page size though? What's the native filesystem page size > in your situation? > I suspect having a single row per page (or worse, fewer than 1 per page) is > not ideal perf-wise, no?
Yes, the small page size was chosen for other reasons, I think we have some work to do to find a better size here. But that's not really germane to this, I just wanted to understand why the -shm grew so large, and now I understand the maths better. Thanks, - Kim _______________________________________________ sqlite-users mailing list [email protected] http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
