v19 : FILLFACTOR and tablelevel AUTOVACUUM recommendations, background worker analysis

[jobinau]

• FILLFACTOR and AUTOVACUUM recommendations per table wherever relevant.
• Additional analysis query to generate FILLFACTOR and AUTOVACUUM-related ALTER statements
• Metadata update for new minor version releases
• Table bloat will be displayed in a separate column as an integer. Fixing the sorting problem
• Full query string as a tooltip on mouseover
• Query will be dumbed into console log
• AWK script enhanced for removal of more comment lines from the source. This helps to add more detailed comments in the source code
• Complete the Rewrite of the blocking sessions analysis. The first step towards deprecation of legacy query:

  pg_gather/gather.sql

  Line 180 in 9adf2a1

  COPY (SELECT blocked_locks.pid AS blocked_pid,

  This also fixes a problem where victims are listed as part of blockers. Now blocker is identified separately and highlighted Victims are marked separately. Associated blockers is also mentioned in the mouseover.
• BufferPin wait event doc
• Response time analysis and inference based o historical info
• End-of-life check for PG versions
• WAL compression check for PG 15+
• Eliminate CheckpointWriteDelay and PgSleep from database time
• List biggest consumers of maintenance_work_mem in findings
• blockers and victims analysis query and summary data.
• Data collection time and WAL generation rate in the report
• Record the time at the end of execution in pg_gather_end

Bug Fix:

• Wait event doc reference is cleaned up due to a bug in the Awk script. Make sure that the removed comment don't have any non-whitespace characters before the comment
• Fix for javascript error when there is no blocking sessions. the victim's section of the JSON comes as null
• Fix too-low threshold recommendations. There is not much point in setting a value less tha 500, which will be set as a lower line of recommendation
• Fix: loopholes while collecting and analyzing the wait events of the background process because there are main loops waits which should be considered as idle.
• Fix : Fix count pg_toast_temp as part of temporary schema
• Fix : Handle empty string in workmem and summary calculation
• Fix : Missing table as part of v18 development (pg_get_pidblock)

---

This discussion was created from the release v19 : FILLFACTOR and tablelevel AUTOVACUUM recommendations, background worker analysis.