Fix index out of bounds issue in batch_remove (deepseek-ai#256)

Author: SF-Zhou

src/storage/chunk_engine/src/core/engine.rs

@@ -657,26 +657,26 @@ impl Engine {
         max_count: u64,
     ) -> Result<u64> {
         let chunks = self.meta_store.query_chunks(begin, end, max_count)?;
-        let chunks_len = chunks.len();
+        let mut offset = 0;
+
         const BATCH_SIZE: Size = Size::mebibyte(1);
         let mut write_batch = RocksDB::new_write_batch();
-        let mut persist_end = 0;
         for (index, (chunk_id, meta)) in chunks.iter().enumerate() {
             if write_batch.size_in_bytes() >= BATCH_SIZE.0 as _ {
                 self.meta_store.write(write_batch, true)?;
-                for i in persist_end..index + 1 {
-                    self.meta_cache.remove(&chunks[i].0);
-                }
-                persist_end = index + 1;
                 write_batch = RocksDB::new_write_batch();
+                for (chunk_id, _) in &chunks[offset..index] {
+                    self.meta_cache.remove(chunk_id);
+                }
+                offset = index;
             }
             self.meta_store
                 .remove_mut(&chunk_id, &meta, &mut write_batch)?;
         }
         if !write_batch.is_empty() {
             self.meta_store.write(write_batch, true)?;
-            for i in persist_end..chunks_len {
-                self.meta_cache.remove(&chunks[i].0);
+            for (chunk_id, _) in &chunks[offset..] {
+                self.meta_cache.remove(chunk_id);
             }
         }
         Ok(chunks.len() as _)