Making sense of Slack error logging

[ikesau]

We have two Slack logging functions. Both can be configured to log to a particular Slack channel via an environment variable containing a Slack webhook URL. As far as I know, only prod has these webhook environment variables set.

logContentErrorAndMaybeSendToSlack

• Logs to #content-errors
• Missing title, slug or content for key insight (WordPress)
• No valid key insights found within block (WordPress)
• Title or anchor missing for key insights block (WordPress)
• No fallback title found for prominent link (WordPress)
• Missing data value
• A self redirect (/a -> /a) has been detected
• A circular redirect (/a -> /b -> /a) has been detected

logErrorAndMaybeSendToSlack

• Logs to #developers-errors-server
• GitCMS writeFile failure
• GitCMS deleteFile failure
• renderFrontPage failure
• Google Docs baking failure
• tryDeploy failure
• bake failure
• bakeAndDeploy failure
• apiRouter.delete("/datasets/:datasetId" failure
• apiRouter.put("/datasets/:datasetId" failure

This thread exists for a discussion as to whether or not this strategy is serving us well.

I get the sense that logging to Slack should only be done when an error occurs that requires fixing, i.e. an error has occurred that has broken a page or the site in some way.

Should we look into making the errors more specific or having more channels? Would it be desirable to have the errors tag relevant groups on Slack? Recently it seems that charts have been baked with missing data and it's taken people browsing the site to find them. Maybe some extra logging would be able to prevent that.

Or, are we actually 99% okay with the current state of Slack error logging and nothing needs to be done?

[ikesau]

Outmoded by #2210