Py-qgis-server has two cache for Qgis projects: a LRU cache and a Static cache.
The lru cache store projects and use a lru (last-recent-use) eviction scheme. This cache has a fixed size controlled by the CACHE_SIZE configuration setting.
The lru cache will prevent to bloat the memory with too many projects (remember that projects are loaded in memory for each worker).
If you have many project that are accessed frequently then you may experience many eviction/reloading. This may be not desirable with big projects that may take long loading time, in this situation you may consider using the static cache.
The static cache is not subject to LRU eviction and has no limitation in the number of project you may store.
Projects stored in the static cache are preloaded at startup from a configuration file. The file must have one project uri per line. Each uri is similar to the project uri passed in the ‘MAP’ query parameter of OWS requests.
The path of the cache configuration file is given in the CACHE_PRELOAD_CONFIG configuration setting.
Cache may be checked for invalidation/refresh synchronously or asynchronously depending on the value of the CACHE_CHECK_INTERVAL configuration setting.
If the refresh interval value is set to a strict positive value (>0) then the cache is checked for invalidation/refresh asynchronously every seconds set by the option’s value.
If the refresh interval is set to a negative or null value (<=0) then the cache is invalidated/refreshed synchronously at each requests.
Depending of the backend storage and the loading time of your projects you may choose one or another invalidation strategy.
With slow loading projects it is recommended to use asynchronous check in conjunction with static_cache.