Add TestReplayXversion module for cross-version WAL replay testing#42
Open
x4m wants to merge 7 commits into
Open
Add TestReplayXversion module for cross-version WAL replay testing#42x4m wants to merge 7 commits into
x4m wants to merge 7 commits into
Conversation
Test that WAL generated by the .0 release of a major version replays correctly on the current STABLE binary. This catches backwards- compatibility regressions in WAL replay code, such as the self-deadlock in RecordNewMultiXact introduced by commit 0852643e1c6. The .0 binary is built once and cached. Each run generates fresh WAL (including 2500 multixacts via the savepoint trick to cross an SLRU page boundary) and verifies that a STABLE standby can replay it.
Run the .0 regression tests via pg_regress against the .0 primary for diverse WAL coverage. The regression test files are preserved in inst-dot0/regress/ during the build phase. A 180-second watchdog kills pg_regress if it hangs.
The .0 regression tests run against the .0 server, so there are no feature mismatches and allow_in_place_tablespaces is unnecessary. Use -m immediate for pg_ctl stop so a deadlocked standby does not hang the test indefinitely.
A STABLE branch is created months before the GA release, so the REL_x_0 tag may not exist yet. Silently skip rather than reporting a build failure.
- Set allow_in_place_tablespaces on the STABLE standby so it can replay in-place tablespace WAL from .0 regression tests. - Add -t 10 timeout and SIGKILL fallback to stop_and_clean so a deadlocked standby does not hang the test indefinitely. - Reduce --max-concurrent-tests to 4 to keep checkpoint time short. - Remove unused fields (major, pgsql) and File::Basename import.
Author
|
We use autotools for .0 build. I'm not sure about it, can swap to meson if needed. $stable_inst is done via buildfarm pipeline. Maybe I'll try to switch build of .0 to buildfarm pipeline too. |
When the module runs, add replay_xversion to PG_TEST_EXTRA so that future TAP tests in src/test/recovery/ can be gated on this token. Harmless until PostgreSQL adds such tests.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Test that WAL generated by the .0 release of a major version replays correctly on the current STABLE binary. This catches backwards- compatibility regressions in WAL replay code, such as the self-deadlock in RecordNewMultiXact introduced by commit 0852643e1c6.
The .0 binary is built once and cached. Each run generates fresh WAL (including 2500 multixacts via the savepoint trick to cross an SLRU page boundary) and verifies that a STABLE standby can replay it.
Written using Cursor, needs more testing. I've fixed some issues that catched my eye, but some more review is definitely needed.