csun5285 opened a new pull request, #70763:
URL: https://github.com/apache/doris/pull/70763
### Background
Eight `INVISIBLE` session variables override omitted `VARIANT` column
properties (`max_subcolumns_count`, `enable_typed_paths_to_sparse`,
`max_sparse_column_statistics_size`, `sparse_hash_shard_count`,
`enable_doc_mode`, `doc_materialization_min_rows`, `doc_hash_shard_count`,
`enable_nested_group`) during DDL parsing. PR #70501 made them test-only: their
checker rejects `SET` unless `enable_debug_points` is enabled, and
`LogicalPlanBuilder` reads them only when `enable_debug_points` is enabled.
### What problem does this PR solve?
Issue Number: None
Related PR: #70501 (made the variables test-only)
PR Type: Bugfix
Problem Summary:
After #70501, upgraded clusters whose persisted metadata holds a non-default
`default_variant_*` value—such as Doris 3.1's built-in
`default_variant_max_subcolumns_count = 0` (changed to `2048` in 4.0) or a
pre-upgrade `SET GLOBAL default_variant_*`—propagate that value into every new
session, causing `UNSET VARIABLE ALL` and `UNSET GLOBAL VARIABLE ALL` to fail
without `enable_debug_points` because resetting the value back to the built-in
default runs through the rejected checker. The `default_variant_*` prefix also
does not indicate that these variables are test-only. This PR renames all eight
variables to `debug_variant_*` (and switches `test_index_disk_usage_tvf` to set
`variant_enable_doc_mode = false` in the column definition); 131 of the 137
changed files are regression test renames. After an upgrade, old
`default_variant_*` entries in images and edit logs are ignored so sessions
keep built-in values and `UNSET [GLOBAL] VARIABLE ALL` succeeds, while
explicitly setting an
old `default_variant_*` name fails as an unknown system variable.
### Details
- **Change map:** 137 files changed (`+374/-337`): 3 FE production files
(`+26/-26`), 3 FE unit test files (`+53/-14`), and 131 regression test files
(`+295/-297`).
- FE production (`SessionVariable.java`, `LogicalPlanBuilder.java`,
`PropertyAnalyzer.java`, `+26/-26`): rename the eight variable names and
constants from `default_variant_*` / `DEFAULT_VARIANT_*` to `debug_variant_*` /
`DEBUG_VARIANT_*` (keeping Java field and getter names unchanged), update the
checker error message and comments, and drop the `or turn off
default_variant_enable_nested_group` hint from the escaped-table NestedGroup
error in `PropertyAnalyzer`.
- FE unit tests (`VariableMgrTest.java`, `PropertyAnalyzerTest.java`,
`TypeTest.java`, `+53/-14`): update variable names in existing tests and add
`testOldDefaultVariantNamesIgnored` covering image JSON load, edit-log replay,
and `SET` of an old name.
- Regression tests (131 `.groovy` files, `+295/-297`): replace
`default_variant_*` with `debug_variant_*` across 130 suites, and change
`test_index_disk_usage_tvf` to specify `PROPERTIES("variant_enable_doc_mode" =
"false")` on the `VARIANT` column instead of setting session variables.
- **Frequency:** Every `UNSET VARIABLE ALL` / `UNSET GLOBAL VARIABLE ALL` on
an upgraded cluster whose image or edit log holds a non-default
`default_variant_*` global value (such as any cluster upgraded from 3.1), and
`test_index_disk_usage_tvf` whenever a pipeline runs without
`enable_debug_points`.
- **Root cause:** `SessionVariable` persisted global variable values under
the same `default_variant_*` names across versions. When
`VariableMgr.setAllVarsToDefaultValue` (`UNSET [GLOBAL] VARIABLE ALL`) saw a
value differing from the built-in default, it attempted to write the built-in
default back through `SessionVariable.checkDefaultVariantSettable`, which
rejects all writes when `Config.enable_debug_points` is `false`.
- **Why existing tests missed it:** Master CI pipelines run with
`enable_debug_points = true` and start from fresh metadata, so neither
pre-upgrade persisted `default_variant_*` values nor debug-points-off `UNSET
VARIABLE ALL` / `test_index_disk_usage_tvf` were exercised.
- **Fix:** Rename all eight variables from `default_variant_*` to
`debug_variant_*`. Unknown keys in persisted images
(`SessionVariable.readFromJson`) and edit logs
(`VariableMgr.replayGlobalVariableV2`) are ignored, so upgraded clusters keep
built-in values for `debug_variant_*` and `UNSET [GLOBAL] VARIABLE ALL` skips
them because the current value already equals the default. Update
`test_index_disk_usage_tvf` to set `variant_enable_doc_mode = false` in the
`VARIANT` column definition.
- **Other paths and branches:** Checked all references to `default_variant_`
/ `DEFAULT_VARIANT_` across `fe/`, `be/`, and `regression-test/`; none remain
in FE or regression suites outside the new upgrade test in `VariableMgrTest`.
BE constants (`DEFAULT_VARIANT_MAX_SPARSE_COLUMN_STATS_SIZE`,
`DEFAULT_VARIANT_SPARSE_HASH_SHARD_COUNT`) are BE-internal defaults unrelated
to session variables. A follow-up pick to `branch-4.2` will be submitted
separately.
- **Behavior change:** The eight test-only variables are renamed from
`default_variant_*` to `debug_variant_*`. Setting an old `default_variant_*`
name now fails with `Unknown system variable` (suggesting `debug_variant_*`),
and persisted pre-upgrade `default_variant_*` values in images/edit logs are
ignored on upgrade.
### Testing
- **FE build & checkstyle (local):** `build.sh --fe` succeeded (`BUILD
SUCCESS`, 0 checkstyle violations).
- **FE UT (local):** `run-fe-ut.sh --run
org.apache.doris.qe.VariableMgrTest,org.apache.doris.common.PropertyAnalyzerTest,org.apache.doris.catalog.TypeTest,org.apache.doris.nereids.parser.NereidsParserTest,org.apache.doris.catalog.CreateTableLikeTest,org.apache.doris.catalog.VariantEscapeTableTest`
— 6 classes, 191/191 passed (`VariableMgrTest` 19, `PropertyAnalyzerTest` 35,
`TypeTest` 22, `NereidsParserTest` 92, `CreateTableLikeTest` 14,
`VariantEscapeTableTest` 9), including
`testDebugVariantVariablesOnlyForTesting`,
`testPersistedDebugVariantVariableDroppedOnReplay`, and
`testOldDefaultVariantNamesIgnored`.
- **4.1 -> upgraded FE test with production config (local,
`enable_debug_points = false`, `use_fuzzy_session_variable = false`):**
- Started a Doris 4.1 FE, ran `SET GLOBAL
default_variant_max_subcolumns_count = 0` and `SET GLOBAL
default_variant_enable_doc_mode = true`, waited until persisted in checkpoint
`image.25`, then ran `SET GLOBAL default_variant_sparse_hash_shard_count = 3`
(edit-log-only) and stopped the 4.1 FE.
- Started the newly built FE on the 4.1 `doris-meta` without
`enable_debug_points` or fuzzy variables; verified 4.1 databases loaded and
edit-log replay ignored `default_variant_sparse_hash_shard_count`.
- Verified via `mysql`: `UNSET VARIABLE ALL` and `UNSET GLOBAL VARIABLE
ALL` both succeed; `SET` / `SET GLOBAL` / `UNSET VARIABLE` / `SET_VAR` on
`debug_variant_*` are rejected (`only for testing`); `SET` / `SET GLOBAL` /
`UNSET VARIABLE` / `SET_VAR` on `default_variant_*` fail as unknown system
variables; `SHOW VARIABLES LIKE 'debug_variant_%'` and `'default_variant_%'`
are empty.
- **Regression tests (local, debug points on + fuzzy):** 14 suites passed
covering all 8 renamed variables, `SET_VAR` hints, `test_index_disk_usage_tvf`,
`test_variant_create_table_like`, `test_variant_search_subcolumn`, `rqg7`, and
`check_before_quit`.
### Release note
The eight test-only `VARIANT` session variables (`default_variant_*`) are
renamed to `debug_variant_*`. Setting a `default_variant_*` variable now fails
with `Unknown system variable`. Upgraded clusters with previously persisted
`default_variant_*` global values ignore those old keys and no longer fail on
`UNSET VARIABLE ALL` / `UNSET GLOBAL VARIABLE ALL`.
### Check List (For Author)
- Behavior changed: Yes (see Details)
- Does this need documentation: No (the variables are `INVISIBLE` and
test-only; they are not documented)
- Pre-submission status: Ready
### AI Disclosure
- Tool: Claude Code and Antigravity
- Generated: code edits, unit test updates, regression test renames, and
this description
- Human review: Directed the final design (pure rename without adding a
`VarAttrDef` flag, old names error as unknown variable, column-level property
in `test_index_disk_usage_tvf`) and reviewed the changes.
<!-- Recommended: develop the change with the develop-with-evidence skill
and write this description with the doris-pr-description skill, both from
https://github.com/apache/doris-skills (the doris-skills submodule of this
repository). -->
### Background
<!-- How the feature or module this PR touches works today, in two to four
plain sentences, for a reviewer who doesn't know it. Link the user doc if there
is one. Leave the problem and the fix to Problem Summary. -->
### What problem does this PR solve?
<!-- Issue Number: "close #N" if this PR resolves the issue, "#N" if it only
relates to it, "None" if there is none. A Feature needs an issue. -->
<!-- Related PR: "#N (relation)" for each PR a reviewer should know about,
such as "#123 (introduced the bug)", separated by commas. Write "None" if no
related PR is known. -->
<!-- PR Type: Bugfix, Performance, Feature, Enhancement, Refactor, Test,
Build/CI/Chore, Revert, Release/Dependency, or Other. -->
Issue Number:
Related PR:
PR Type:
Problem Summary:
<!-- Two to five sentences: the problem or need and who is affected; the
approach in one sentence; when the diff is larger than that suggests, what else
it had to handle, or that most of it is tests; and what existing users see
after an upgrade. -->
### Details
<!--
For a diff of more than about 300 changed lines, start with "- **Change
map:**": how many lines are production code, tests, generated files, and docs,
then the code in parts, each with the problem it handles and its size.
Then answer each field of your PR type on its own line, as "- **Field:**
answer", in this order:
Bugfix: Frequency, Root cause, Why existing tests missed it, Fix, Other
paths and branches
Performance: Why it was slow, Optimization, Results, Applicability and cost
Feature: Scope, Design, Usage, Default behavior, Compatibility
Enhancement: Before and after, Configs and defaults, Impact on existing
users
Refactor: Moves vs. logic changes, Behavior change, Compatibility,
Performance impact, Rollback
Test, adding tests: Scenarios, What existing tests miss, Fails when broken
Test, fixing a test: Failure, Cause, Stability evidence
Build/CI/Chore: Affected, Developer and release impact, License and
third-party, Rollback
Revert: Why not fix forward, Original author notified, Follow-up
Release/Dependency: Versions, Upstream changes, Compatibility, Rollback
Other: Why Other
"None" is a valid answer when it is true. Don't restate the diff.
-->
### Testing
<!-- The tests, suites, or commands that ran, what each shows, and where it
ran (locally or in CI). For a fix, the test that fails before it and passes
after. -->
### Notes for Reviewers
<!-- Optional; delete this section if there is nothing to say. Why each
non-obvious edit was made, what you are unsure about, and where to start
reading if Details has no Change map. -->
### Release note
<!-- One paragraph on what users will notice, or None. -->
### Check List (For Author)
<!-- Behavior changed: "No", or "Yes (see Details)" with a Behavior change
field in Details. -->
<!-- Does this need documentation: "No", or "Yes" with the
apache/doris-website PR or the plan. -->
<!-- Pre-submission status: "Ready"; "Partially verified: " and what is not
verified; or "Not ready: " and why, for a Draft PR. -->
- Behavior changed:
- Does this need documentation:
- Pre-submission status:
### AI Disclosure
<!-- Tool: the agent and model, or None. Generated: the parts it wrote
(code, tests, this description), or None. Human review: what you reviewed of
the generated parts, or "Not applicable" without AI tools. You may add "-
Prompts:" to summarize how you directed the agent. -->
- Tool:
- Generated:
- Human review:
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]