Odoo operations · Technical reference
Deployment, upgrades, backups and the operational work that keeps an instance alive.
24 entries · free, no sign-up
Showing all 24 entries
Odoo.sh is ideal for rapid development and built-in CI/CD. For 500+ users, heavy processing, or strict data residency, dedicated AWS/GCP with Kubernetes is preferred for better scaling control.
Nginx acts as a Reverse Proxy handling SSL termination, static file caching, request buffering, and load balancing between workers.
It handles real-time communications (Chat, Notifications) via /websocket. Nginx must route these requests specifically to the gevent-based 8072 port.
Monitor dead tuples and tune autovacuum aggressively. For extreme cases, use pg_repack to reclaim space without locking tables.
It's a connection pooler that allows hundreds of Odoo workers to share a smaller pool of Postgres connections, reducing DB overhead.
Analyze EXPLAIN ANALYZE output. Usually solved by adding missing B-tree or GIN indexes, or refactoring ORM calls to avoid search() loops.
Use SSH keys, disable /web/database/manager, implement Fail2Ban, run Odoo as a dedicated system user, and keep Postgres on a private subnet.
Use Environment Variables or Secret Managers (AWS Secrets, Vault) rather than hardcoding. Avoid storing secrets in the database where possible.
Used for DB management via UI. In production, it should be a strong hash in odoo.conf (admin_passwd) and the /web/database selector should be disabled.
Use tools like Prometheus with a PostgreSQL exporter, or Odoo-specific monitors like 'Odoo Performance Monitor'. For errors, integrate Sentry to capture Python tracebacks and JS errors.
Odoo can generate GBs of logs daily. Use 'logrotate' to compress and delete old logs, preventing the disk from filling up and crashing the server.
Monitor the RSS (Resident Set Size) of worker processes. If they grow indefinitely, it's a sign of a leak. Odoo's 'limit_memory_soft' helps by recycling workers after they reach a certain threshold.
It's a way for the server to push data to the browser. It uses a gevent-based server on port 8072 because long-held connections would block standard workers.
Check pg_hba.conf for permissions, verify the db_host and db_port in odoo.conf, and check for 'Too many connections' in Postgres logs.
While Odoo doesn't natively support read/write splitting, you can use a middleware like 'Pgpool-II' to route read queries to replicas and write queries to the primary.
Use Amazon S3 (with a custom module) or a mounted NFS/EFS volume. All app servers must access the same /var/lib/odoo/filestore to see attachments/images.
A coroutine-based Python networking library. Odoo uses it for the 'Long-polling' server to handle thousands of concurrent connections efficiently.
Horizontal scaling: Multiple Nginx load balancers -> 10+ Odoo App Servers -> Large RDS/Postgres Cluster -> Pgbouncer -> Redis for session storage.
An architecture where app servers don't store local data. Sessions are in Redis, files are in S3, and the database is remote. This allows servers to be destroyed/re-created easily.
Every GitHub branch is a separate Odoo instance. 'Production' is for live, 'Staging' for testing with real data, and 'Development' for unit tests.
A web-based terminal that allows you to run odoo-bin commands, inspect logs, or interact with the database directly in the cloud environment.
Go to the 'Backups' tab and click 'Import' or 'Restore' on a specific branch. It's automated and handles both DB and filestore.
Odoo 19 introduces stricter scopes for API keys, allowing you to limit a key to specific models or methods (e.g., only 'read' on 'sale.order').
By using a faster internal bundler and better caching for OWL components, reducing the time it takes for the first page load in production.
Next
We run technical assessment for Odoo hires, and we take on the work these questions describe.