timescaledb
timescaledb copied to clipboard
Forcefully enable tiered data for CAGGs
When cagg refresh occurs and is triggered by the background worker tiered data may be excluded if timescaledb.enable_tiered_reads
is off on the server level. Also when refresh is triggered manually by a user whether tiered data appears in the cagg or not depends on the user's session settings despite of the server settings.
This PR fixes that by forcefully enabling the timescaledb.enable_tiered_reads
GUC in the cagg refresh procedure.
I don't think it can be done without having the osm extension installed. I'm planning to add these tests in osm. But also after having a chat with @gayyappan we might want to rework this PR, gotta discuss these changes further.
...or maybe it can be done with file_fdw
or postgres_fdw
, will check.
...or maybe it can be done with
file_fdw
orpostgres_fdw
, will check.
We have tests with fdw setup for testing osm. see chunk_utils_internal.sql.
Codecov Report
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 80.77%. Comparing base (
59f50f2
) to head (57f62a8
). Report is 154 commits behind head on main.
Additional details and impacted files
@@ Coverage Diff @@
## main #5351 +/- ##
==========================================
+ Coverage 80.06% 80.77% +0.70%
==========================================
Files 190 199 +9
Lines 37181 37234 +53
Branches 9450 9732 +282
==========================================
+ Hits 29770 30075 +305
- Misses 2997 3200 +203
+ Partials 4414 3959 -455
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Updating this old PR (thought it was merged). Added a test demonstrating that tiered data can be accessed when refreshing CAggs.
Updating this old PR (thought it was merged). Added a test demonstrating that tiered data can be accessed when refreshing CAggs. Didn't we decide not to go ahead with this?