hcq/oceanbase
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[CP] fix lost row caused by purge range

Browse Source
This commit is contained in:
kongfy
2022-04-25 14:20:04 +08:00
committed by wangzelin.wzl
parent 80133e10bc
commit 8ce40d33af
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/storage/ob_multiple_scan_merge_impl.cpp
View File

@ -75,6 +75,7 @@ int ObMultipleScanMergeImpl::supply_consume()
for (int64_t i = 0; OB_SUCC(ret) && i < consume_num; ++i) {
const int64_t iter_idx = consumer_.get_consumer_iters()[i];
ObStoreRowIterator* iter = iters_.at(iter_idx);
item.iter_idx_ = iter_idx;
if (NULL == iter) {
ret = common::OB_ERR_UNEXPECTED;
STORAGE_LOG(WARN, "Unexpected error", K(ret), K(iter));
@ -88,7 +89,6 @@ int ObMultipleScanMergeImpl::supply_consume()
ret = common::OB_ERR_UNEXPECTED;
STORAGE_LOG(WARN, "get next row return NULL row", "iter_index", iter_idx, K(ret));
} else {
item.iter_idx_ = iter_idx;
if (try_push_top_item_) {
if (1 != consume_num) {
ret = OB_ERR_UNEXPECTED;
@ -104,8 +104,8 @@ int ObMultipleScanMergeImpl::supply_consume()
// TODO: Ambiguous here, typically base_row only means row in major sstable.
// And iter_idx==0 doesn't necessarily mean iterator for memtable.
0 == iter_idx ? ++row_stat_.inc_row_count_ : ++row_stat_.base_row_count_;
range_purger_.on_push((int)item.iter_idx_, item.iter_flag_);
}
range_purger_.on_push((int)item.iter_idx_, item.iter_flag_);
}
if (OB_SUCC(ret)) {