Workers & Performance

import { Aside } from ‘@astrojs/starlight/components’;

Odoo uses a multi-process architecture with workers to handle concurrent requests. Properly configuring workers is key to good performance.

What Are Workers?

Odoo workers are separate processes that handle incoming HTTP requests. More workers means more concurrent requests can be processed simultaneously.

• 0 workers — Single-threaded mode (development only)
• 1 worker — Handles one request at a time, suitable for light usage
• 2+ workers — Handles multiple concurrent requests

A general rule: allocate **1 worker per 6 concurrent users** and ensure you have at least **1 GB RAM per worker**.

Worker Configuration

Recommended Settings by Usage

Usage	Concurrent Users	Workers	RAM	CPU
Development	1-2	1	2 GB	1 vCPU
Small team	5-10	2	4 GB	2 vCPU
Medium team	10-30	4	8 GB	4 vCPU
Large team	30-100	8	16 GB	8 vCPU

Changing Workers

1. Go to your instance → Settings → Performance
2. Adjust the worker count
3. Click Save — Deploy Monkey restarts the instance with the new configuration

Plan Limits

Plan	Max Workers
Free	1
Base	2
Pro	Unlimited
Agency	Unlimited

Cron Workers

Odoo uses a separate cron worker to execute scheduled actions (automated emails, inventory checks, etc.). Deploy Monkey allocates:

• 1 cron worker by default for all instances
• This worker does not count toward your worker limit

Memory Limits

Each worker has memory limits to prevent runaway processes:

Setting	Default	Description
limit_memory_soft	2 GB	Worker is recycled after exceeding this
limit_memory_hard	2.5 GB	Worker is killed immediately
limit_time_cpu	60s	Max CPU time per request
limit_time_real	120s	Max wall-clock time per request

If your workers are frequently being recycled (check the logs), you may need to increase memory limits or optimize your custom modules.

Worker Usage Monitoring

Deploy Monkey monitors your worker processes in real-time to help you understand when scaling is needed.

What’s Measured

During regular server health checks (every 5 minutes), Deploy Monkey:

1. Identifies all Odoo/Python worker processes in your container
2. Measures CPU usage over a 0.5-second window
3. Classifies workers as busy (consumed CPU) or idle
4. Calculates usage percentage: busy workers ÷ configured workers × 100

Viewing Worker Data

Worker usage appears in two places:

1. Instance Monitoring tab — worker usage chart alongside health metrics
2. Worker Usage API — GET /api/v1/instances/{id}/worker-usage for programmatic access

Scaling Recommendations

Usage Level	Recommendation
< 60%	Normal — no action needed
60-80%	Moderate — monitor during peak hours
> 80%	High pressure — consider adding workers
> 95%	Critical — users may experience slow responses

Worker usage is an estimate based on process activity sampling. Actual concurrency may vary depending on request types (long-running reports vs quick page loads).

Performance Tips

Database Optimization

• Vacuum regularly — Odoo runs auto-vacuum, but large databases benefit from manual vacuum during maintenance windows
• Index custom fields — If you query custom fields frequently, add database indexes
• Archive old records — Move old sales orders, leads, and logs to archive

Odoo Configuration

• Enable proxy mode — Already configured by Deploy Monkey (proxy_mode = True)
• Tune db_maxconn — Database connections per worker. Default is 64, which works for most setups
• Use --load=base,web — Only load necessary server-wide modules

Server-Level

• Use SSDs — Odoo is I/O intensive; SSDs make a significant difference
• Adequate RAM — Underprovised RAM leads to swap usage, which kills performance
• Fast CPU — Single-thread performance matters more than core count for Odoo

Application-Level

• Minimize custom module complexity — Avoid heavy computations in compute fields
• Use read_group for reporting — Instead of loading all records and computing in Python
• Batch operations — Use write on recordsets instead of looping with individual writes

Longpolling

Odoo uses longpolling for real-time features like chat and notifications. Deploy Monkey configures a dedicated longpolling port and routes it through Nginx automatically.

The longpolling process:

• Runs on a separate port (default 8072)
• Handles WebSocket-like connections for real-time updates
• Is proxied through Nginx at /longpolling/poll
• Does not count toward your worker limit