From d1c11bcf0896d20a2a0454b7db9d420aaf1c7987 Mon Sep 17 00:00:00 2001 From: Elias Holzmann <9659253+EliasHolzmann@users.noreply.github.com> Date: Sun, 19 Oct 2025 01:01:28 +0200 Subject: [PATCH] Documentation: correct info re PAPERLESS_THREADS_PER_WORKER (#11031) --- docs/configuration.md | 19 ++++--------------- 1 file changed, 4 insertions(+), 15 deletions(-) diff --git a/docs/configuration.md b/docs/configuration.md index 61e82363e..51c00d1a5 100644 --- a/docs/configuration.md +++ b/docs/configuration.md @@ -980,21 +980,10 @@ paperless will process in parallel on a single document. process very large documents faster, use a higher thread per worker count. - The default is a balance between the two, according to your CPU core - count, with a slight favor towards threads per worker: - - | CPU core count | Workers | Threads | - | -------------- | ------- | ------- | - | > 1 | > 1 | > 1 | - | > 2 | > 2 | > 1 | - | > 4 | > 2 | > 2 | - | > 6 | > 2 | > 3 | - | > 8 | > 2 | > 4 | - | > 12 | > 3 | > 4 | - | > 16 | > 4 | > 4 | - - If you only specify PAPERLESS_TASK_WORKERS, paperless will adjust - PAPERLESS_THREADS_PER_WORKER automatically. + If unset, paperless uses `max(floor(cpu_count / PAPERLESS_TASK_WORKERS), 1)` + threads per worker. The idea behind this is that as long as there are enough cores, + the total number of threads should less than or equal to the total number of (logical) + CPU cores. #### [`PAPERLESS_WORKER_TIMEOUT=`](#PAPERLESS_WORKER_TIMEOUT) {#PAPERLESS_WORKER_TIMEOUT}