GetAllPretranslations and other GetAll* handlers under src/Serval/src/Serval.Translation/Features/Engines/ and src/Serval/src/Serval.WordAlignment/ materialize the entire result set into memory and JSON (potentially millions of pretranslation rows). A single request can drive memory pressure / OOM and long GC pauses.
Fix: Add server-enforced pagination (page-size cap, e.g. 1000) and stream results (IAsyncEnumerable) rather than buffering whole lists.
GetAllPretranslationsand otherGetAll*handlers undersrc/Serval/src/Serval.Translation/Features/Engines/andsrc/Serval/src/Serval.WordAlignment/materialize the entire result set into memory and JSON (potentially millions of pretranslation rows). A single request can drive memory pressure / OOM and long GC pauses.Fix: Add server-enforced pagination (page-size cap, e.g. 1000) and stream results (
IAsyncEnumerable) rather than buffering whole lists.