Skip to content

Code Insights Still In Queue After Deleting From Dashboard

Created by: alexAtSourcegraph

  • Sourcegraph version: 3.38.0
  • Platform information: Kubernetes

Steps to reproduce:

  • Create the following insights:
    • select:repo file:yarn.lock archived:no fork:no
    • try {:[_]} catch (:[e]) {} finally {:[_]} lang:java patternType:structural archived:no fork:no
  • Delete both insights from the dashboard
  • Check the primary database to see if the jobs are in any other state rather than terminated with the following query
    • select count(*) from insights_query_runner_jobs where state in ('queued', 'processing', 'errored')

Expected behavior:

After deleting all the insights from the dashboard, running the following query should return a count of 0 and the jobs should be terminated.

select count(*) from insights_query_runner_jobs where state in ('queued', 'processing', 'errored')

Actual behavior:

After deleting all insights from the dashboard, searcher evictions occur every 30 - 60 minutes and there is a queue of insights_query_runner_jobs in various states other than terminated.