• v3.3.45 79fe3037eb

    Administrator released this 2026-06-07 20:41:25 +02:00 | 10 commits to master since this release

    Hotfix for the freeze on Start with thousands of queued jobs. The 3.3.43 diagnostic logging probed every queued job's file head and emitted an 'upload-start' rot-log event BEFORE the semaphore, which fired 3000× simultaneously when a big batch started — 3000 fs.open syscalls + 3000 main→renderer IPC messages back-to-back, blocking the main process and the renderer event loop. Probe and rot-log now run AFTER semaphore acquisition (only the currently-uploading parallel-count jobs do the work). The duplicate fs.statSync per job is gone (reuses fileSize from the existing pre-check). Diagnostic rot-log events ('upload-start', 'upload-failure', 'hoster-transient', 'fast-fail', etc.) are still written to account-rotation.log and to the per-job log buffer, but no longer forwarded to the renderer — only user-visible rotation events (switchAccount, mark-failed, doodstream-via-api, …) reach the toast handler now.

    Downloads