Fix incorrect index->reltuples after VACUUM #217
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Given the scenario of updating stats during VACUUM: 1) coordinator vacuums and updates stats of its own; 2) then coordinator dispatches vacuum to segments; 3) coordinator combines stats received from segments to overwrite the stats of its own. Because upstream introduced a feature which could skip full index scan uring cleanup of B-tree indexes when possible (refer to: postgres/postgres@857f9c3), there was a case in QD-QEs distributed deployment that some QEs could skip full index scan and stop updating statistics, resulting in QD being unable to collect all QEs' stats thus overwrote a paritial accumulated value to index->reltuples. More interesting, it usually happened starting from the 3rd time of consecutively VACUUM after fresh inserts due to above skipping index scan criteria. This patch is intended to prevent from above issue by two aspects: on QE, do not skip full index scan, report current statistics to QD as requested; on QD, restrict updating statistics only when collecting all QEs' data completely, if the reporting QE number doesn't match the total number of dispatched QEs, no stats update happens.
|
|
lss602726449
force-pushed
the
vaccum_stat
branch
from
00c7c43 to
fd157d5
Compare
my-ship-it
previously approved these changes
Sep 22, 2023
avamingli
previously approved these changes
Sep 22, 2023
There is conflict in _bt_vacuum_needs_cleanup between CloudberryDB and GPDB. We update the _bt_vacuum_needs_cleanup in CloudberryDB. We enforce update stats when Gp_role is GP_ROLE_EXECUTE, and set estimated_count to false. which in the scan_index We call vac_update_relstats to send back stats to QE. Fix serveral tests due to update stats after vacuum. We keep these tests result same with GPDB.
lss602726449
dismissed stale reviews from avamingli and my-ship-it
via
28060dd
lss602726449
force-pushed
the
vaccum_stat
branch
from
fd157d5 to
28060dd
Compare
avamingli
approved these changes
Sep 25, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
closes: #214
Change logs
We enforce all QE to update stats after vacuum in the index, and all QD will collect stats and update the pg_class. Otherwises, only some QE will update stats and send back to QD which left index stats in incomplete state.
Why are the changes needed?
Describe why the changes are necessary.
Does this PR introduce any user-facing change?
no
How was this patch tested?
cherrypick
Contributor's Checklist
Here are some reminders and checklists before/when submitting your pull request, please check them:
make installcheckmake -C src/test installcheck-cbdb-parallel