In the webinar we'll walk you through the new functionality that helps you find the right settings for autovacuum scale factor and threshold (to avoid bloat), detect surprising issues related to xmin horizon (aka "Dead Tuples Not Yet Removable"), new in-depth metrics related to freezing and autovacuum scheduling, and more.
Looking back, the PGCon talk recordings have been uploaded to YouTube, including our own talk on Index Selection using Constraint Programming. Listen in to learn more about the new constraint programming model that will power the next generation of Index Advisor in pganalyze.
We've also made a few other updates to pganalyze that you can find below (including a handful of Index Advisor bug fixes), and as always you can find our most recent episodes of "5mins of Postgres" in this newsletter.
In this product webinar on July 26th, 9:30am PT, you will learn about the most important VACUUM metrics, concepts like xmin horizon, how to prevent transaction ID wraparound, estimating bloat, and more.
We will walk you through the upcoming pganalyze VACUUM advisor, its per-table tuning recommendations for autovacuum settings, and how the new functionality was built over the last year.
We're talking about about the growing pains of database architecture at Figma as well as alternate architectures, such as the sharding architecture chosen by Notion.
We talk about forcing join order in Postgres. We walk through 3 techniques: Adding OFFSET “0” to a sub-SELECT, using common table expressions, and working with the join_collapse_limit.
We explain what the Translation Lookaside Buffer is and how, starting with Postgres 15, Postgres can help you calculate how many huge pages to allocate.
Subscribe on YouTube
We publish weekly videos on Postgres, our conference talks, tutorials, product walk-throughs, and more.
Add support for analyzing queries that require the "ltree" extension
Fix analysis error when foreign key constraint tables have indexes using custom functions
Fix INSERT INTO SELECT errors by ensuring correct column order for query analysis
This fixes type coercion errors like "column "X" is of type A but expression is of type B" on specific cases of INSERT INTO SELECT statements that relied on the column positions of the target table.
Add support for IMMUTABLE functions
This fixes errors like "functions in index expression must be marked IMMUTABLE" or "functions in partition key expression must be marked IMMUTABLE".
Fix some cases of query analysis error "view depth limit (10) exceeded"
This fixes an edge case that caused Index Advisor to fail for particular queries, due to a table and a view having the exact name, but different schemas, with the view referencing the table of the same name.
Query Performance
Adjust estimation method used for calculate overview percentile metrics
This may cause small differences in top-level percentile numbers going forward compared to historic calculations
Other changes
Config Settings: Update and sync config groups with Postgres 15
Connection Tracing: Increase Connection Traces live refresh timeout from 1 to 5 minutes
Check-Up: Show full query text instead of truncated one-line query for issue pages of Active Queries and Blocking Queries checks
Query details page: Update query text unavailable error message to explain how to fix the problem queries
Snapshot API: Require collector API keys, do not accept read or read/write keys
Review us on G2 or TrustRadius
Please leave your feedback (good or bad!) in this G2 survey or in the TrustRadius survey below.
