csun5285 opened a new pull request, #70742:
URL: https://github.com/apache/doris/pull/70742
…
### What problem does this PR solve?
Issue Number: None
Related PR: #70462, #69515
Problem Summary:
A query of a DUP_KEYS table neither merges nor folds rows: it reads its
rowsets one after another, and each rowset reader reads its segments one after
another. It still went through VCollectIterator, which is built for merging: a
Level1Iterator over one Level0Iterator per rowset, an IteratorRowRef cursor,
and a first Block read eagerly inside BlockReader::init() to seed that cursor.
This is part of the series that moves one read mode at a time off
VCollectIterator (#70462 moves the AGG_KEYS fold reads).
- BlockReader::_unions_rowsets() decides once in init() both the setup and
the next-block function: a DUP_KEYS query (READER_QUERY) without
read_orderby_key, which is every DUP_KEYS query but a key TopN and a changelog
MIN_DELTA or DETAIL scan. These are the scans the FE sends in direct mode.
Compactions, schema change, checksum and every read of the other key types keep
VCollectIterator.
- The new mutable BE config enable_new_block_iterator (default true, random
in fuzzy mode), the one #70462 adds for the AGG_KEYS reads, turns the union on;
when it is false every DUP_KEYS query reads through VCollectIterator as before.
A reader reads it once, in init().
- BlockReader::_init_rowsets_union() captures and inits the rowset readers
as _init_collect_iter() does, forwards a pushed-down LIMIT to each as
VCollectIterator::add_child() does, and puts them in a VUnionIterator, each
behind a RowsetReaderIterator adapter. Nothing is read in init(); a rowset
reader opens its segments on its first next_batch(). VUnionIterator releases a
rowset reader, and with it the rowset, once it is read.
- vgeneric_iterators.h gets the #pragma once it never had: block_reader.cpp
now includes it, and the unity build puts block_reader.cpp and
vgeneric_iterators.cpp into one translation unit.
- OlapScanner::_get_block_impl() appends the tablet and backend to read
errors, as _open_impl() does for init() (#64755): the first Block of a DUP_KEYS
query is now read there.
### Release note
None
### Check List (For Author)
- Test: Unit Test, Regression test
- New BlockReaderRowsetsUnionTest (4 cases; the row-order and LIMIT
cases run with enable_new_block_iterator on and off and expect the same rows);
read-path and compaction unit tests (BlockReader*, VGenericIterators*,
*Changelog*, *Compaction*, SegCompaction*, *RowIdConversion*,
TabletReaderTest*, *SchemaChange*, Vertical*, RowTtl*, VCollect*, OlapScanner*,
BetaRowset*, *SegmentIterator*): 795 run, 792 pass, 3 skip themselves (ASAN).
- New regression suite fault_injection_p0/test_dup_query_rowsets_union
(.out generated), which pins enable_new_block_iterator to true.
- Regression on a local ASAN cluster: the new suite,
test_index_compound_directory_fault_injection, data_model_p0/duplicate,
delete_p0, changelog_p0, table_stream_p0 and topn_optimize, 101/101 with
enable_new_block_iterator on and 100/100 with it off (the new suite left out).
- Before the rebase and the config, on the same read path: query_p0,
inverted_index_p0, fault_injection_p0, 1376 suites: 1341 pass, 6 more pass on a
rerun, 29 fail for reasons of the machine (object storage, sshd, Python UDF,
plugin, missing FE config, a p1 query timeout under ASAN, a compaction-delay
assertion); details in the PR description.
- Behavior changed: Yes. Read errors of OLAP scans end with ". tablet=<id>,
backend=<host>" like init() errors. With enable_new_block_iterator on (the
default), for DUP_KEYS queries the first Block is read in the first get_block()
instead of init(), so its time moves from ReaderInitTime to the scan time, and
BlockReaderVcollectIterInitTimer / BlockReaderBuildHeapInitTimer stay 0. Query
results do not change. New mutable BE config enable_new_block_iterator (default
true); false keeps the read path of master.
- Does this need documentation: No
<!-- 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]