There have been a number of test/format failures timing out with cache stuck full in the last few days. The failures generally have a set of shared characteristics:
- LSM
- Moderately sized cache (30MB->60MB)
- No transaction IDs are pinned
An example config from a failing run:
############################################ # RUN PARAMETERS ############################################ abort=0 alter=0 auto_throttle=1 backups=0 bitcnt=4 bloom=1 bloom_bit_count=40 bloom_hash_count=25 bloom_oldest=0 cache=120 checkpoints=1 checksum=uncompressed chunk_size=4 compaction=0 compression=lz4 data_extend=0 data_source=lsm delete_pct=5 dictionary=0 direct_io=0 encryption=none evict_max=2 file_type=row-store firstfit=1 huffman_key=0 huffman_value=0 independent_thread_rng=1 in_memory=0 insert_pct=69 internal_key_truncation=1 internal_page_max=9 isolation=random key_gap=5 key_max=256 key_min=256 leaf_page_max=9 leak_memory=0 logging=0 logging_archive=1 logging_compression=none logging_prealloc=1 long_running_txn=0 lsm_worker_threads=4 merge_max=8 mmap=1 modify_pct=0 ops=100000 prefix_compression=1 prefix_compression_min=3 quiet=1 read_pct=7 rebalance=1 repeat_data_pct=42 reverse=0 rows=100000 runs=1 salvage=1 split_pct=73 statistics=0 statistics_server=0 threads=19 timer=360 transaction_timestamps=0 transaction-frequency=52 value_max=2671 value_min=256 verify=1 wiredtiger_config= write_pct=19 ############################################
A link to the original failure:
http://build.wiredtiger.com:8080/job/wiredtiger-test-format-stress/56914
- related to
-
WT-3616 format failed to report a stuck cache
-
- Closed
-
- links to