From 3e4ff4019bd864ec4bd305db34c76f83f63d2588 Mon Sep 17 00:00:00 2001
From: Davide Scaini <davide.scaini@alumni.cern>
Date: Fri, 10 Apr 2026 18:13:49 +0200
Subject: [PATCH] limit number of workers

---
 bincio/extract/cli.py    | 2 +-
 bincio/extract/config.py | 2 ++
 docs/deployment/vps.md   | 2 ++
 3 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/bincio/extract/cli.py b/bincio/extract/cli.py
index b95ec37..6ddf179 100644
--- a/bincio/extract/cli.py
+++ b/bincio/extract/cli.py
@@ -171,7 +171,7 @@ def extract(
     dedup = DedupIndex(output_dir=cfg.output_dir)
     known_hashes: frozenset = frozenset(dedup._by_hash.keys())
 
-    n_workers = workers or os.cpu_count() or 4
+    n_workers = workers or cfg.workers or os.cpu_count() or 4
     console.print(f"Using [bold]{n_workers}[/bold] worker processes.")
 
     owner = {"handle": cfg.owner_handle, "display_name": cfg.owner_display_name}
diff --git a/bincio/extract/config.py b/bincio/extract/config.py
index 1c8c6a1..baa72d6 100644
--- a/bincio/extract/config.py
+++ b/bincio/extract/config.py
@@ -51,6 +51,7 @@ class ExtractConfig:
     track: TrackConfig = field(default_factory=TrackConfig)
     classifier: ClassifierConfig = field(default_factory=ClassifierConfig)
     incremental: bool = True
+    workers: Optional[int] = None   # None → use CPU count
     owner_handle: str = "me"
     owner_display_name: str = "Me"
     athlete: AthleteConfig | None = None
@@ -109,6 +110,7 @@ def load_config(path: Path) -> ExtractConfig:
         track=track,
         classifier=classifier,
         incremental=raw.get("incremental", True),
+        workers=raw.get("workers"),
         owner_handle=owner.get("handle", "me"),
         owner_display_name=owner.get("display_name", "Me"),
         athlete=athlete,
diff --git a/docs/deployment/vps.md b/docs/deployment/vps.md
index 4c1e905..1a5f9b5 100644
--- a/docs/deployment/vps.md
+++ b/docs/deployment/vps.md
@@ -206,6 +206,8 @@ sources:
 
 output:
   dir: /var/bincio/data
+
+workers: 2   # cap extract parallelism on the VPS (default: all CPUs)
 ```
 
 Sync and extract (run from your laptop or SSH in):