[fix](case) Relax show data compaction size check

[shuke987]

Summary

• Relax strict SHOW DATA size equality in test_show_data_with_compaction.
• Assert the compacted documents table size falls into the observed stable physical range for cloud/non-cloud runs.
• Keep the small insert table comparison as a relative size check after full compaction.

Testing

• git diff --check -- regression-test/suites/inverted_index_p2/test_show_data.groovy
• Remote branch-4.1 regression single suite before final range-only adjustment: ./run-regression-test.sh --run -d inverted_index_p2 -s test_show_data_with_compaction -parallel 1
  • Result: All suites success. Test 1 suites, failed 0 suites, fatal 0 scripts, skipped 0 scripts

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[shuke987]

run buildall

[shuke987]

/review

[github-actions]

Reviewed PR 65387.

I did not find a correctness issue worth an inline review comment. The patch only relaxes physical SHOW DATA size equality in test_show_data_with_compaction to a bounded relative comparison. The compared values are still checked for wait_timeout before the helper is called, full compaction is still explicitly awaited, and nearby inverted-index show-data tests already use relative tolerance for physical size comparisons.

Validation was static/GitHub-patch based in this runner: the checkout is shallow, local base lookup for the prompt base SHA is incomplete, and I did not run the p2 regression suite.

User focus: no additional user-provided review focus was supplied.

Subagent conclusions: optimizer-rewrite reported no optimizer/rewrite, semantic equivalence, join, aggregate, or parallel-path issue. tests-session-config reported no regression-test, expected-output, session/config propagation, compatibility, or style issue. Final convergence round 1 ended with both live subagents replying NO_NEW_VALUABLE_FINDINGS for the same empty inline-comment set.

[shuke987]

run buildall

[shuke987]

/review

[github-actions]

Reviewed PR 65387 at 736406c.

I did not find a correctness issue worth an inline review comment.

Critical checkpoints:

• Changed-file scope: GitHub reports one modified file, regression-test/suites/inverted_index_p2/test_show_data.groovy.
• Existing review context: the prepared inline-thread context has no existing inline review comments or replies, so there was no duplicate inline issue to suppress.
• Test behavior: the patch only relaxes SHOW DATA physical-size assertions in test_show_data_with_compaction; the table definitions, stream loads, sync, full-compaction waits, inverted_index_compaction_enable save/restore path, and SQL/query semantics are unchanged.
• Expected outputs: this suite has no .out artifact to update.
• User focus: no additional user-provided review focus was supplied.
• Validation limits: review was static/GitHub-patch based in this runner. The checkout does not contain the prompt base SHA, and thirdparty/installed, thirdparty/installed/bin/protoc, and built output/ artifacts are missing, so I did not run the p2 regression suite locally.

Subagent conclusions:

• optimizer-rewrite found no optimizer/rewrite, semantic-equivalence, join, aggregate, or parallel-path issue.
• tests-session-config found no regression-test, expected-output, session/config propagation, compatibility, generated-output, or basic CI/style issue.
• Final convergence round 1 ended with both live subagents replying NO_NEW_VALUABLE_FINDINGS for the same empty proposed inline-comment set.

Repository review skill note: I searched the checkout for a repo-local SKILL.md / code-review skill before reviewing, but no such file was present in this runner, so I followed the explicit .code-review.ECQuo4/review_prompt.txt workflow.

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

PR approved by anyone and no changes requested.

[shuke987]

skip buildall

[shuke987]

/review

[shuke987]

Address review feedback by keeping the relative size comparison between the documents table paths while preserving the cloud/non-cloud sanity ranges.\n\nrun buildall

[github-actions]

Automated review summary for PR #65387.

I reviewed the single changed file, regression-test/suites/inverted_index_p2/test_show_data.groovy, including the surrounding test_show_data_with_compaction flow, the SHOW DATA output shape, compaction helper behavior, cloud-mode split, and comparable inverted-index size/compaction tests.

Opinion: I found no blocking issue and no substantiated inline comment to submit. The change keeps the documents-table enabled/disabled compaction comparison, adds absolute range checks for the observed cloud/non-cloud SHOW DATA sizes, and relaxes both equality checks into a bounded relative-difference assertion.

Critical checkpoints:

• Review prompt, prepared review context, changed-file list, existing review threads/comments, user focus, required-agents file, and shared ledger were read before diff review.
• No AGENTS.md files were required for the changed-file ancestors.
• The expected repo-local .claude/skills/code-review/SKILL.md was not present after hidden-path search, so I followed the explicit review prompt and loaded workflow contract.
• GitHub PR metadata confirmed head a490d6e00261113c5d745f55e2d020cb2e1d5923 is still current and the PR scope is exactly one modified file.
• Existing inline review context was empty, so there were no prior threads to suppress.
• GitHub patch whitespace/trailing-space checks found no issue. Local ancestry-based git diff --check was not usable because this checkout is shallow, so I used GitHub patch metadata for scope and validation.
• I did not run the full regression suite locally in this runner.

Subagent conclusions:

• optimizer-rewrite: no optimizer/rewrite, semantic-equivalence, or parallel join/aggregate issue found.
• tests-session-config: no regression-test, expected-output, session/config propagation, compatibility, or basic CI/style issue found.
• Convergence round 1 ended with both live subagents replying NO_NEW_VALUABLE_FINDINGS for the same ledger state and empty proposed inline comment set.

User focus: no additional user-provided review focus was supplied.

[shuke987]

skip buildall