Pre-launch checklist

[jpmckinney]

OCP:

• YL: Check for missing publications. Create an issue for Jan to start the processing of those datasets. Add recent publications #92
• UI text
  • JM/SZ: Check UI text (skipped – defer to comms team)
  • JM: Ask Sophie A and Georg for copy-writing
  • SA/GN: Do copy-editing
  • YL: Coordinate translation of UI text to Spanish and Russian Russian translation #97
• Dataset descriptions
  • JM: Read English dataset descriptions for content (skipped - defer to comms team)
  • JM: Ask Sophie A and Georg for copy-writing
  • SA/GN: Do copy-editing
  • YL: Assign the preparation of quality summaries to helpdesk analysts
  • YL: Assign Spanish and Russian translations after copy-writing is done
  • JM/YL: Fill in final text for all datasets and check metadata
  • JM/YL: Put “We have not yet prepared a data quality summary for this dataset.” if missing
  • JM: Decide what to do about licensing, since we found no lawyers to help (sheet)
• Application testing
  • Test the registry-spoonbill integration, especially with files that would exceed Excel's limits
  • SZ to run through all screens, to check we didn't miss something
• Deployment
  • Delete /var/www/.htpasswd (basic authentication)
  • Remove symlinks /data/kingfisher-collect and /data/logs once new env files are deployed

Quintagroup:

• Deploy Spoonbill to the production server, so that it is ready for integration by Datlab

Datlab/OCP:

• Activate scheduled runs of all active sources (@yolile can identify those that have stopped publishing)
• @hrubyjan said some collections might need to be re-processed, because they were processed with older code
  • Before re-processing, check for a completed job first, since it just might not be active (e.g. Colombia needs the date ranges, etc. to be filled in, due to a migration error).
• Implement feedback email
• Check for any other blockers to launch

Dogsbody:

• Set up data.open-contracting.org
• Review open deploy issues tagged 'registry'
• Discuss whether to switch from gunicorn to uWSGI (used in all other apps)
  • JM: I think gunicorn makes more sense inside Docker, since uWSGI requires a lot of configuration.
• VACUUM FULL or similar to reduce size of DB Registry: Monitor DB size and VACUUM FULL if needed deploy#290
• Backup the data registry database and the export dumps: Registry: Read-only backup server deploy#289
• Suggestion: for data registry, hot backup in read-only mode, low CPU, etc. requirements, just need large enough drives – need to ask Quinta about Spoonbill: Registry: Read-only backup server deploy#289

Post-launch:

• JM: Add data.open-contracting.org and flatten.open-contracting.org to Google Webmaster Central and Google Search Console

@hrubyjan Please add anything that is missing.

[jpmckinney]

Test the registry-spoonbill integration, especially with files that would exceed Excel's limits

#180

Activate scheduled runs of all active sources

Management command has a cron job. Related: #181

Add data.open-contracting.org and flatten.open-contracting.org to Google Webmaster Central and Google Search Console

open-contracting/deploy#336