Hi! I have a few questions/observations about the caching model. Any help clarifying these issues would be greatly appreciated. (I'm working on a Windows Server class OS).
1) Setting the SQLITE_CONFIG_PAGECACHE keyword within sqlite3_config() seems to define a single cache for the instance of the database. However, from what I've observed/read, each connection to the database has its own cache (Unless the shared cache mode is enabled). So how does the single buffer defined by SQLITE_CONFIG_PAGECACHE relate to the cache for each connection instance? 2) I had stupidly defined the SQLITE_DEFAULT_PAGE_SIZE to be 4096, and the SQLITE_DEFAULT_CACHE_SIZE to 200,000 within a program that has 15+ database connections. When I started hammering my program, I saw the memory shoot up to well over 1.5GB and continue climbing. I came to realize that each connection had a maximum cache size of 750MB which combined would easily exceed the 2GB program limit. What's interesting is that after I let the app sit with no work for about an hour and then executed a single write and commit, the app size shrunk by about 1GB which implies the cache was cleaned out. Does anybody have details on when and how the cache is flushed? Thanks, Mike Borland _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
