Skip to content

Latest commit

 

History

History
74 lines (57 loc) · 3.15 KB

production-instructions.md

File metadata and controls

74 lines (57 loc) · 3.15 KB
Raw

Use of Uwazi in production environment

SaaS

Software as a service is offered in https://www.uwazi.io/.

On your server (Uwazi official documentation)

Please look at the official documentation (in English) at huridocs/uwazi Wiki on https://github.com/huridocs/uwazi/wiki/Install-Uwazi-on-your-server.

For other languages, see https://uwazi.io/page/9852italrtk/support.

With uwazi-docker

In short: uwazi-docker is not designed to run in production. If in the future enough volunteers with both experience and usage on production environment feel more confident, this status could change. Follow these instructions at your own risk. Suggestions are welcome.

Where is my data?

Your data is stored on docker volumes. Each volume can be attached to only one active container. The most reliable way to backup is to take a snapshot of all data containers at the same time, so consider this when doing a backup in an active moment of your application.

The data volumes (and sources of data if you are using Uwazi without docker) that contain real data that cannot be recreated are:

  • MongoDB database
  • uploaded_documents folder
  • Elastic Search database (Cache that needs to be recreated)

You can backup and restore the Elastic Search database; still, it is possible to recreate its indexes by running yarn reindex from a non-dockerized uwazi or a uwazi-docker dockerised uwazi service; it is basically a cache for search. At this current version uwazi-docker does not automate yarn reindex, so you would need to use docker exec to do it once after restoring a backup.

To see the names of the docker volumes, use this command:

docker volume ls | grep uwazi

This will show something similar to:

local     uwazi-docker_elasticsearch_data1
local     uwazi-docker_mongodb_data1
local     uwazi-docker_uploaded_documents

How to backup docker volumes

See docker official documentation on Backup, restore, or migrate data volumes. This link also shows how to move data from docker volume to your host filesystem.

See blacklabelops/volumerize for a powerfull automated solution for periodic backups.

Did you have extra tips or improvements that could make uwazi-docker a bit more friendly to run on production but still be able to be tested for unskilled people? If you are willing to release your collaboration under public domain, you can submit suggestions in https://github.com/fititnt/uwazi-docker/issues.