Skip to content

qBittorrent Performance Tuning

Understanding the Settings

qBittorrent has two types of limits that are often confused:

Queue Limits (which torrents can seek peers)

  • max_active_uploads: How many seeding torrents can be "active" (allowed to find/connect peers)
  • max_active_downloads: How many downloading torrents can be "active"
  • max_active_torrents: Total active torrents (uploads + downloads)

Critical: If max_active_uploads is too low, torrents get stuck in queuedUP state and cannot even attempt to connect to peers. This is the #1 cause of slow speeds with many seeding torrents.

Bandwidth/Slot Limits (how many can transfer data)

  • max_uploads: How many torrents can upload data simultaneously
  • max_uploads_per_torrent: Upload slots per torrent
  • max_connec: Total peer connections across all torrents
  • max_connec_per_torrent: Peer connections per torrent

Memory Limits

  • Container memory limit: Too low causes OOMKilled restarts, disrupting all connections
  • Each torrent + peers consumes memory; 2000+ torrents need 4-6 GiB

Symptoms and Root Causes

Symptom: Many torrents "Stalled Uploading" or "queuedUP"

Root cause: max_active_uploads too low

Torrents in queuedUP state cannot connect to peers at all. They're waiting in a queue for an "active" slot.

Example: With 2000 seeding torrents and max_active_uploads=100: - Only 100 torrents can seek peers - 1900 torrents stuck in queue, never connecting - Total peers: ~100-300 (very low)

Fix: Increase max_active_uploads to 400-600 (25-30% of seeding count)

Symptom: Slow downloads despite many seeds

Root causes: 1. max_active_downloads too low (torrents queued, not connecting) 2. Upload blocked → BitTorrent tit-for-tat penalizes you 3. Container restarting (OOMKilled) disrupts connections

BitTorrent protocol: Peers prioritize uploaders. If you're not uploading, you get slower download slots.

Symptom: Container restarting frequently

Root cause: Memory limit too low

Check with: 

kubectl describe pod qbittorrent-0 -n qbittorrent | grep -A5 "Last State"

If "OOMKilled", increase memory limit in values.yaml.

Symptom: High SYN_SENT:CLOSED on router (>500)

Root cause: max_connec too low

Peers try to connect but qbittorrent can't accept them (connection slots full). Connections stay stuck at SYN_SENT then timeout.

Fix: Increase max_connec based on router capacity (see below).

Router Capacity Analysis

pfSense state table stores each network connection. Each torrent peer creates ~19 state entries over its lifecycle (SYN → ESTABLISHED → FIN_WAIT → TIME_WAIT).

Check Capacity

ssh pfsense.grigri "pfctl -s info | grep -E 'current entries'"
ssh pfsense.grigri "pfctl -s memory"  # shows limit

Calculate Safe Limits

Router State Limit Safe Ceiling (50%) Max Connections
402,000 (default) 201,000 states ~10,500 connections
200,000 100,000 states ~5,250 connections

Formula: max_connec = (safe_ceiling_states - baseline_states) ÷ 19

Where baseline_states = states from other services (~2-5k)

Example Calculation

Router limit: 402,000
Safe ceiling: 201,000 (50%)
Current non-torrent states: 5,000
Available for torrents: 196,000
Max connections safe: 196,000 ÷ 19 = ~10,000
Recommended: 8,000 (leaving buffer)

Quick Diagnostics

Check Queue State

curl -s http://qbittorrent.internal.grigri.cloud/api/v2/sync/maindata | jq '
  .torrents | to_entries |
  group_by(.value.state) |
  map({state: .[0].value.state, count: length})
'

Key states to watch: - queuedUP: Stuck waiting for active slot (BAD) - stalledUP: Active but no peers connected (OK if low) - uploading: Actively transferring (GOOD)

Check Container Health

kubectl --context=grigri describe pod qbittorrent-0 -n qbittorrent | grep -A3 "Last State"
kubectl --context=grigri top pod qbittorrent-0 -n qbittorrent --containers

Check Router State

ssh pfsense.grigri "pfctl -s states | grep 50413 | grep -c ESTABLISHED"
ssh pfsense.grigri "pfctl -s states | grep -c 'SYN_SENT:CLOSED'"

Key Metrics

Metric Healthy Warning Critical
queuedUP torrents < 5% 5-20% > 20%
stalledUP torrents < 20% 20-40% > 40%
Total peer connections 500+ < 200 < 50
SYN_SENT:CLOSED < 100 100-500 > 500
Router state % < 25% 25-50% > 80%
Container restarts/day 0 1-3 > 5

For High-Seeding Scenarios (2000+ torrents)

Setting Recommended Purpose
max_active_uploads 400-600 Queue slots - allows torrents to seek peers
max_active_downloads 10-20 Queue slots for downloads
max_active_torrents 600-800 Total queue limit
max_uploads 60-100 Bandwidth slots - actual upload transfers
max_uploads_per_torrent 4-6 Per-torrent upload slots
max_connec 5000-8000 Total peer connections (router dependent)
max_connec_per_torrent 20-25 Per-torrent peers
Memory limit 6 GiB Prevent OOMKilled

Relationship Summary

seeding_torrents ≈ 2000
max_active_uploads ≈ 25-30% of seeding = 500-600
max_uploads ≈ bandwidth_based (9 MB/s ÷ 90 KB/s/slot = 100)
max_connec ≈ router_safe (8000 with 400k limit)

Tuning Workflow

  1. Diagnose the bottleneck:
  2. High queuedUP → increase max_active_uploads
  3. High SYN_SENT:CLOSED → increase max_connec
  4. OOMKilled → increase memory limit

  5. Slow downloads → check upload is working

  6. Adjust via API: 

    curl -X POST http://qbittorrent.internal.grigri.cloud/api/v2/app/setPreferences \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "json={'max_active_uploads':600,'max_connec':8000}"

  7. Verify router capacity: 

    ssh pfsense.grigri "pfctl -s info | grep 'current entries'"

  8. Monitor 10-30 minutes:

  9. queuedUP should drop to < 5%
  10. Peer connections should increase
  11. Upload/download speeds should improve

Common Pitfalls

Pitfall Result Fix
max_active_uploads = 100 with 2000 seeding 1900 torrents queued Increase to 500+
max_connec = 500 with 2000 seeding SYN_SENT overflow Increase to 5000+
Memory = 3.5 GiB with 2000 torrents OOMKilled restarts Increase to 6 GiB
max_uploads_per_torrent = 1 Single torrent hogs bandwidth Use 4-6

Quick Fixes

All torrents stuck in queue

# Increase active slots
curl -X POST http://qbittorrent.internal.grigri.cloud/api/v2/app/setPreferences \
  -d "json={'max_active_uploads':600,'max_active_torrents':800}"

Slow downloads

# Ensure uploads work (tit-for-tat)
curl -s http://qbittorrent.internal.grigri.cloud/api/v2/transfer/info | jq '.up_info_speed'
# If 0 or very low, check max_uploads_per_torrent isn't too low

Router SYN_SENT high

# Increase connection limit
curl -X POST http://qbittorrent.internal.grigri.cloud/api/v2/app/setPreferences \
  -d "json={'max_connec':8000}"

Container restarting

# Check OOMKilled
kubectl describe pod qbittorrent-0 -n qbittorrent | grep OOMKilled
# If present, increase memory limit in values.yaml and redeploy