2021.06.0 Release

Docker tag: quay.io/pganalyze/enterprise:v2021.06.0

Notes

  • This release requires a database migration when coming from older releases, be sure to follow the recommended upgrade steps
  • The query fingerprint format has been changed due to the pg_query 2.0 update - historic data will show as separate query records from the query records for new data

New features

  • Update to pg_query 2.0 and Postgres 13 query parser
  • Integrate with Slack for alerts about new issues
    • To use the new Slack integration, create your own Slack application for use with pganalyze, and then configure SLACK_CLIENT_ID and SLACK_CLIENT_SECRET
  • Integrate with PagerDuty for alerts about new critical issues
    • To use the new PagerDuty integration, create your own PagerDuty application for use with pganalyze, and then configure the PAGERDUTY_APP_ID - reach out to support for more details
  • Introduce server-wide issues
    • Previously certain issues showed up multiple times, once for each database on a server, e.g. when a configuration setting should be tuned
    • This is now clearly separated into a server-wide list, with database-specific issues shown separately
    • If you use the public API method getIssues we recommend querying by serverId instead going forward
  • New issue detail page
    • This gives an overview of an issue triggered by a check, shows its activity, describes what triggered it and links to the check configuration, and offers guidance on how to resolve the problem.
    • E-mail alerts for critical issues now link to this issue detail page, and in-app issue lists link to the page as well.
  • Use Postgres scanner for syntax highlighting
    • This handles Postgres-specific keywords better when SQL text is shown in-app (e.g. on the query details page)
    • In addition, table references within a query are now linked to the table details page.
  • Re-activate weekly reports
    • This was previously turned off for performance reasons, but has now been refactored and improved
    • If MAILER_URL is set on your installation all organization members will start receiving weekly report emails every Monday
    • This can be turned off in each user's profile settings (accessible through the link on the bottom left of the navigation)
  • Re-activate unused index check
    • This was previously turned off for performance reasons, but is now active again with updated logic
    • Unused indexes are identified as indexes that have not had any scans for 2 weeks
  • Members: Add "copy invite link" to invite new members without requiring MAILER_URL to be set
  • Add support for AWS WebIdentity for use in Amazon EKS
  • Per-table VACUUM graph: Full support for 7-day history, improve visual cues
  • Per-table ANALYZE: Show related configuration settings

Bugfixes

  • DEFAULT_ORG_ROLE now defaults to none, ensuring all users that sign up need an invite to join an organization
  • Show relation-does-not-exist errors in log-based EXPLAIN errors (instead of "Internal Error")
  • Active query / Idle TX check: Only consider backend type "client backend"
  • Index invalid check: Increase threshold for when indices are detected from 1 hour to 24 hours (to better handle long CREATE INDEX CONCURRENTLY builds)
  • Minor fixes to EXPLAIN node formatting:
    • Better handling for long CTE names
    • Include Sort Key field as a significant field
    • Fix array-valued significant fields
  • Replication followers: Don't list followers without any activity
  • Organization-wide database list: Auto-grow server name field
  • Upgrade built-in pganalyze collector from 0.37.0 to 0.40.0
    • Reduce timeout for accessing EC2 instance metadata service
    • Fix reporting of replication statistics for more than 1 follower
    • Amazon RDS: Support long log events beyond 2,000 lines

Performance improvements

  • Per-table query overview: Optimize backend queries for better performance
  • Per-table VACUUM graph: Calculate more data server-side, keeping data sent to client smaller
  • Pre-aggregate daily and hourly schema table/index statistics
  • Top navigation: Avoid N+1 on each server's database list
  • Organization overview: Fix N+1 for loading database counts for each server

Security improvements

  • Update OS base image packages to latest versions
  • Update app dependencies for CVE-2021-28965, CVE-2021-22904, CVE-2021-22885, CVE-2021-22903, CVE-2021-22902, CVE-2021-29509, GHSA-7rrm-v45f-jp64

Other changes

  • Rename all user-facing instances of "indices" to "indexes", to match Postgres project terminology

Couldn't find what you were looking for or want to talk about something specific?
Start a conversation with us →