Example threads. These questions and answers were written by the SDODS team to document real problems and how to fix them. The people named are illustrative, not real users. Questions asked on this site are marked new.
sdods insights show -p checkout-apiNo insights computed yet for checkout-api.
hint: Run `sdods insights compute -p checkout-api` after ingesting some runs.Fine, so I ran compute:
sdods insights compute -p checkout-apiInsights need the @sdods/db package and a configured database.We have run this suite every night for a week and every run is sitting in .sdods/runs with its summary and its ndjson. Why does counting them need a database?
Accepted answer
Because insights are computed over ingested runs, not over directories. A run directory holds one run's artifacts; flakiness, locator fragility and suite health are all "this scenario across the last N runs", and that comparison lives in the database.
So the missing step is ingest, not compute:
sdods db migrate
sdods report ingest --run-id <id> .sdods/runs/<id>/messages.ndjsonand from then on let the run do it:
sdods run -p checkout-api -e staging -l api --ingestYou can backfill the week you already have — re-ingesting the same files is idempotent, so a loop over the run directories is safe to run twice if it dies halfway.
One expectation to set: a scenario needs at least --min-runs runs before it can be flagged, 10 by default. Seven nightlies will compute fine and flag nothing, which looks like it is still broken. It is not.
Wrap commands and output in triple backticks and they will render as a code block.
Checking whether you’re signed in…
Have a different question? Ask it.