This is an automated email from the ASF dual-hosted git repository.
dataroaring pushed a commit to branch branch-3.0
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/branch-3.0 by this push:
new f4b98400783 branch-3.0: [Fix](merge-on-write) Add defensive check
before partial update #44687 (#45082)
f4b98400783 is described below
commit f4b98400783e207ac2e4e25bcb824ddfafb910fb
Author: bobhan1 <bao...@selectdb.com>
AuthorDate: Fri Dec 6 23:15:47 2024 +0800
branch-3.0: [Fix](merge-on-write) Add defensive check before partial update
#44687 (#45082)
pick https://github.com/apache/doris/pull/44687
---
be/src/olap/rowset/segment_v2/segment_writer.cpp | 39 +++++++++++++++++++---
be/src/olap/rowset/segment_v2/segment_writer.h | 1 +
.../rowset/segment_v2/vertical_segment_writer.cpp | 37 ++++++++++++++++++--
.../rowset/segment_v2/vertical_segment_writer.h | 1 +
4 files changed, 71 insertions(+), 7 deletions(-)
diff --git a/be/src/olap/rowset/segment_v2/segment_writer.cpp
b/be/src/olap/rowset/segment_v2/segment_writer.cpp
index 0db06c2de74..2a61eb3f0fe 100644
--- a/be/src/olap/rowset/segment_v2/segment_writer.cpp
+++ b/be/src/olap/rowset/segment_v2/segment_writer.cpp
@@ -519,6 +519,40 @@ Status SegmentWriter::probe_key_for_mow(
return Status::OK();
}
+Status SegmentWriter::partial_update_preconditions_check(size_t row_pos) {
+ if (!_is_mow()) {
+ auto msg = fmt::format(
+ "Can only do partial update on merge-on-write unique table,
but found: "
+ "keys_type={}, _opts.enable_unique_key_merge_on_write={},
tablet_id={}",
+ _tablet_schema->keys_type(),
_opts.enable_unique_key_merge_on_write,
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ if (_opts.rowset_ctx->partial_update_info == nullptr) {
+ auto msg =
+ fmt::format("partial_update_info should not be nullptr, please
check, tablet_id={}",
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ if (!_opts.rowset_ctx->partial_update_info->is_partial_update) {
+ auto msg = fmt::format(
+ "in fixed partial update code, but is_partial_update=false,
please check, "
+ "tablet_id={}",
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ if (row_pos != 0) {
+ auto msg = fmt::format("row_pos should be 0, but found {},
tablet_id={}", row_pos,
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ return Status::OK();
+}
+
// for partial update, we should do following steps to fill content of block:
// 1. set block data to data convertor, and get all key_column's converted
slice
// 2. get pk of input block, and read missing columns
@@ -536,10 +570,7 @@ Status
SegmentWriter::append_block_with_partial_content(const vectorized::Block*
block->columns(),
_tablet_schema->num_key_columns(),
_tablet_schema->num_columns()));
}
- DCHECK(_is_mow());
-
- DCHECK(_opts.rowset_ctx->partial_update_info);
- DCHECK(row_pos == 0);
+ RETURN_IF_ERROR(partial_update_preconditions_check(row_pos));
// find missing column cids
const auto& missing_cids =
_opts.rowset_ctx->partial_update_info->missing_cids;
diff --git a/be/src/olap/rowset/segment_v2/segment_writer.h
b/be/src/olap/rowset/segment_v2/segment_writer.h
index 4b157092221..02ae962a16a 100644
--- a/be/src/olap/rowset/segment_v2/segment_writer.h
+++ b/be/src/olap/rowset/segment_v2/segment_writer.h
@@ -102,6 +102,7 @@ public:
bool& has_default_or_nullable,
std::vector<bool>& use_default_or_null_flag,
PartialUpdateStats& stats);
+ Status partial_update_preconditions_check(size_t row_pos);
Status append_block_with_partial_content(const vectorized::Block* block,
size_t row_pos,
size_t num_rows);
Status append_block_with_variant_subcolumns(vectorized::Block& data);
diff --git a/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
b/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
index 857f6b2f3c7..cdbdcab4791 100644
--- a/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
+++ b/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
@@ -390,6 +390,39 @@ Status VerticalSegmentWriter::_probe_key_for_mow(
return Status::OK();
}
+Status VerticalSegmentWriter::_partial_update_preconditions_check(size_t
row_pos) {
+ if (!_is_mow()) {
+ auto msg = fmt::format(
+ "Can only do partial update on merge-on-write unique table,
but found: "
+ "keys_type={}, _opts.enable_unique_key_merge_on_write={},
tablet_id={}",
+ _tablet_schema->keys_type(),
_opts.enable_unique_key_merge_on_write,
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ if (_opts.rowset_ctx->partial_update_info == nullptr) {
+ auto msg =
+ fmt::format("partial_update_info should not be nullptr, please
check, tablet_id={}",
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ if (!_opts.rowset_ctx->partial_update_info->is_partial_update) {
+ auto msg = fmt::format(
+ "in partial update code, but is_partial_update=false, please
check, tablet_id={}",
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ if (row_pos != 0) {
+ auto msg = fmt::format("row_pos should be 0, but found {},
tablet_id={}", row_pos,
+ _tablet->tablet_id());
+ DCHECK(false) << msg;
+ return Status::InternalError<false>(msg);
+ }
+ return Status::OK();
+}
+
// for partial update, we should do following steps to fill content of block:
// 1. set block data to data convertor, and get all key_column's converted
slice
// 2. get pk of input block, and read missing columns
@@ -399,9 +432,7 @@ Status VerticalSegmentWriter::_probe_key_for_mow(
// 3. set columns to data convertor and then write all columns
Status VerticalSegmentWriter::_append_block_with_partial_content(RowsInBlock&
data,
vectorized::Block& full_block) {
- DCHECK(_is_mow());
- DCHECK(_opts.rowset_ctx->partial_update_info != nullptr);
- DCHECK(data.row_pos == 0);
+ RETURN_IF_ERROR(_partial_update_preconditions_check(data.row_pos));
// create full block and fill with input columns
full_block = _tablet_schema->create_block();
diff --git a/be/src/olap/rowset/segment_v2/vertical_segment_writer.h
b/be/src/olap/rowset/segment_v2/vertical_segment_writer.h
index 0e04b49b343..a327873bbce 100644
--- a/be/src/olap/rowset/segment_v2/vertical_segment_writer.h
+++ b/be/src/olap/rowset/segment_v2/vertical_segment_writer.h
@@ -173,6 +173,7 @@ private:
bool& has_default_or_nullable,
std::vector<bool>& use_default_or_null_flag,
PartialUpdateStats& stats);
+ Status _partial_update_preconditions_check(size_t row_pos);
Status _append_block_with_partial_content(RowsInBlock& data,
vectorized::Block& full_block);
Status _append_block_with_variant_subcolumns(RowsInBlock& data);
Status _generate_key_index(
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
