Re: Vacuum statistics

Alena Rybakina <a.rybakina@postgrespro.ru>

From: Alena Rybakina <a.rybakina@postgrespro.ru>

To: pgsql-hackers <pgsql-hackers@postgresql.org>

Cc: Alexander Korotkov <aekorotkov@gmail.com>, Amit Kapila <amit.kapila16@gmail.com>, Jim Nasby <jnasby@upgrade.com>, Bertrand Drouvot <bertranddrouvot.pg@gmail.com>, Kirill Reshke <reshkekirill@gmail.com>, Andrei Zubkov <zubkov@moonset.ru>, Masahiko Sawada <sawada.mshk@gmail.com>, Melanie Plageman <melanieplageman@gmail.com>, jian he <jian.universality@gmail.com>, a.lepikhov@postgrespro.ru, Sami Imseih <samimseih@gmail.com>, vignesh C <vignesh21@gmail.com>, Ilia Evdokimov <ilya.evdokimov@tantorlabs.com>, Alena Rybakina <lena.ribackina@yandex.ru>

Date: 2025-12-20T23:36:42Z

Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

Move wal_buffers_full from PgStat_PendingWalStats to WalUsage
- eaf502747bac 18.0 cited

Attachments

v26-0001-Machinery-for-grabbing-an-extended-vacuum-statistics.patch (text/plain) patch v26-0001
v26-0002-Machinery-for-grabbing-an-extended-vacuum-statistics.patch (text/plain) patch v26-0002
v26-0003-Machinery-for-grabbing-an-extended-vacuum-statistics.patch (text/plain) patch v26-0003
v26-0004-Vacuum-statistics-have-been-separated-from-regular.patch (text/plain) patch v26-0004
v26-0005-Add-documentation-about-the-system-views-that-are-us.patch (text/plain) patch v26-0005

Hi,
I’ve added some changes to one of the approaches and also did additional 
cleanup and stabilization work on the vacuum statistics tests. Specifically:

  * I moved the vacuum statistics tests into the tests tab and made them
    more stable. For slower machines, vacuum is now triggered inside the
    statistics wait function. Previously, some backends didn’t have
    enough time to release the lock, which could lead to differences
    because the vacuum hadn’t fully completed yet.
  * I also ran the backend tests and fixed a couple of minor issues
    along the way.
  * I ran pgindent to clean up and normalize the formatting.

For now, I’ve temporarily removed collecting statistics related to 
database-level errors when vacuum is forced to stop. I’m currently stuck 
on how to properly expose statistics for cluster-level objects, since 
their dbid is 0.

At the moment, only the second test still looks odd, and I haven’t fully 
figured out why yet. It seems like aggressive vacuum can no longer be 
triggered the same way as before with the current gucs, but I’m still 
investigating this.

Best regards,
Alena Rybakina