github-actions[bot] commented on code in PR #65058:
URL: https://github.com/apache/doris/pull/65058#discussion_r3542332740


##########
be/src/storage/segment/segment.h:
##########
@@ -215,12 +218,12 @@ class Segment : public 
std::enable_shared_from_this<Segment>, public MetadataAdd
 
     // get the column reader by tablet column, return NOT_FOUND if not found 
reader in this segment
     Status get_column_reader(const TabletColumn& col, 
std::shared_ptr<ColumnReader>* column_reader,
-                             OlapReaderStatistics* stats,
+                             OlapReaderStatistics* stats, const io::IOContext* 
io_ctx = nullptr,

Review Comment:
   Leaving `io_ctx` optional here lets existing query call sites keep bypassing 
the new limiter. For example, `SegmentIterator` still calls 
`get_data_type_of(*col, _opts)`, and `get_data_type_of()` creates the variant 
root reader with `get_column_reader(unique_id, &v_reader, &stats)` instead of 
passing `read_options.io_ctx`. If that is the first touch of an 
external-variant column, `VariantExternalMetaReader::init_from_footer()` loads 
the `variant_meta_keys` indexed column with a null source context, so 
`IndexedColumnReader::read_page()` sees index metadata but no `READER_QUERY` / 
`remote_scan_cache_write_limiter`. `_build_page_zone_maps()` has the same issue 
with `_segment->get_column_reader(*tablet_column, &reader, _opts.stats)`. With 
`file_cache_query_limit_segment_meta=true` and 
`file_cache_query_limit_bytes=0`, these metadata reads can still populate file 
cache. Please make query call sites pass the `StorageReadOptions` IOContext, or 
avoid a nullable default for paths that can r
 un under query reads.



-- 
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]

Reply via email to