Fix CAgg migration performance regression #7517
Open
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
fabriziomello
added
performance
continuous_aggregate
migration
force-auto-backport
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #7517 +/- ##
==========================================
+ Coverage 80.06% 82.22% +2.15%
==========================================
Files 190 231 +41
Lines 37181 43373 +6192
Branches 9450 10909 +1459
==========================================
+ Hits 29770 35663 +5893
- Misses 2997 3383 +386
+ Partials 4414 4327 -87 ☔ View full report in Codecov by Sentry. |
fabriziomello
force-pushed
the
cagg_migrate_fix_performance_regression
branch
12 times, most recently
from
7035bac to
ffedd0a
Compare
|
There is no reason this change should not go in the release notes, so we should add a changelog entry. In addition, this is a non-trivial change, so we should probably stick to two reviewers. |
fabriziomello
force-pushed
the
cagg_migrate_fix_performance_regression
branch
2 times, most recently
from
7ab9690 to
6da1365
Compare
When creating the migration plan we need to figure out the boundaries (min and max) of the primary dimension to split the COPY data into multiple transactions to don't exauste the instance resources. The problem is we're reading the boundaries direct from the CAgg view and when the realtime is enabled it lead to big performance penalty (this was one of the reasons for the new format). Also in the `COPY DATA` migration plan step we changed the batch size used that previously was based on the materialization hypertable partition range and now will be the Continuous Aggregate bucked size multiplied by 10 (ten). Fixed it by querying direct the materialization hypertable instead. Also in ef2cfe3 we introduced a problem that now we're not explicitly updating the watermark but instead warning the user to manually execute the refresh procedure to update it, but without update the watermark the query performance for realtime Continuous Aggregate can be awful so we introduced a small procedure to update the watermark for the new migrated cagg during the migration.
fabriziomello
force-pushed
the
cagg_migrate_fix_performance_regression
branch
from
6da1365 to
04da963
Compare
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.
When creating the migration plan we need to figure out the boundaries (min and max) of the primary dimension to split the COPY data into multiple transactions to don't exauste the instance resources.
The problem is we're reading the boundaries direct from the CAgg view and when the realtime is enabled it lead to big performance penalty (this was one of the reasons for the new format).
Also in the
COPY DATAmigration plan step we changed the batch size used that previously was based on the materialization hypertable partition range and now will be the Continuous Aggregate bucked size multiplied by 10 (ten).Fixed it by querying direct the materialization hypertable instead.
Also in ef2cfe3 we introduced a problem that now we're not explicitly updating the watermark but instead warning the user to manually execute the refresh procedure to update it, but without update the watermark the query performance for realtime Continuous Aggregate can be awful so we introduced a small procedure to update the watermark for the new migrated cagg during the migration.
Disable-check: force-changelog-file