Details
-
Bug
-
Resolution: Unresolved
-
Medium
-
None
-
None
-
3
-
9223372036854775807
Description
llapi_project_get() iterates through the /etc/projid file every time a project id is resolved to a project name to project id or vice versa.
llapi_projid_get() should read the /etc/projid file once build an in-memory cache for the name<->projid mappings, so that repeated llapi_project_getnam() and llapi_project_getprjid() calls can avoid re-scanning the list multiple times. It would also make sense to cache the last PROJID-to-PROJNAME mapping in ll_project_handle to avoid any lookup in the common case of repeated lookup of the same NAME/ID.
If implementation time is a concern, a linear array search would be an acceptable starting point, and still far better than opening and re-reading the file multiple times. Ideally, this would implement a hash table to efficiently look up PROJID and PROJNAME entries, as a large system may have tens of thousands or potentially even millions of projids (eg. if a new project is being automatically created for each Kubernetes container/user or similar).