DBRepair development

[Volts]

I saw a significant drop in performance when pagesize exceeded a certain size. 4096 was an undeniable win. 8192 and 16384 were better only with tons of cache available to SQLite. 32768 and 65536 were worse.

Database Cache Size Behavior Question

And a few more for comparison. Again I acknowledge that this isn’t the same as being inside Plex. This is all speedtest1, new files each time, varying the page_size and cache_size only. # Variants of this command: rm testfile.db ; ./speedtest1 --size 200 --journal WAL --pagesize 1024 --cachesize 2000 testfile.db | grep TOTAL 1024, 2000 = 2048000 TOTAL....................................................... 55.506s 1024, 4000 = 4096000 TOTAL......................…

That was all on a filesystem with the default ZFS recordsize of 128k.