We are seeing a lot of failed stress test runs at the moment because enabling timestamps in test/format often results in the cache being pinned full.
We need to investigate whether we can reasonably avoid the cache being pinned full, and if not update test/format to avoid encountering the condition.
An example of a failure:
[1512735814:255349][22179:0x7f3b7d224700], t, eviction-server: Cache stuck for too long, giving up: Connection timed out 0xe341c0:=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= 0xe341c0:transaction state dump 0xe341c0:current ID: 72453 0xe341c0:last running ID: 72453 0xe341c0:oldest ID: 72453 0xe341c0:commit timestamp: aed7c 0xe341c0:oldest timestamp: ac902 0xe341c0:pinned timestamp: ac902 0xe341c0:stable timestamp: 0 0xe341c0:has_commit_timestamp: yes 0xe341c0:has_oldest_timestamp: yes 0xe341c0:has_pinned_timestamp: yes 0xe341c0:has_stable_timestamp: no 0xe341c0:oldest_is_pinned: yes 0xe341c0:stable_is_pinned: no 0xe341c0:checkpoint running: no 0xe341c0:checkpoint generation: 228 0xe341c0:checkpoint pinned ID: 0 0xe341c0:checkpoint txn ID: 0 0xe341c0:oldest named snapshot ID: 0 0xe341c0:session count: 35 0xe341c0:Transaction state of active sessions: 0xe341c0:=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= 0xe341c0:cache dump 0xe341c0:cache full: no 0xe341c0:cache clean check: no (0%) 0xe341c0:cache dirty check: yes (50%) 0xe341c0:file:wt(<live>): 0xe341c0:internal: 1 pages, 0MB, 0/1 clean/dirty pages, 0/0 clean/dirty MB, 0MB max page, 0MB max dirty page 0xe341c0:leaf: 310 pages, 15MB, 168/142 clean/dirty pages, 9/6 clean/dirty MB, 0MB max page, 0MB max dirty page 0xe341c0:file:WiredTigerLAS.wt(<live>): 0xe341c0:internal: 1 pages, 0MB, 0/1 clean/dirty pages, 0/0 clean/dirty MB, 0MB max page, 0MB max dirty page 0xe341c0:file:WiredTiger.wt(<live>): 0xe341c0:internal: 1 pages, 0MB, 1/0 clean/dirty pages, 0/0 clean/dirty MB, 0MB max page, 0MB max dirty page 0xe341c0:leaf: 1 pages, 0MB, 1/0 clean/dirty pages, 0/0 clean/dirty MB, 0MB max page, 0MB max dirty page 0xe341c0:cache dump: total found: 17MB vs tracked inuse 18MB 0xe341c0:total dirty bytes: 6MB [1512735814:255734][22179:0x7f3b7d224700], t, eviction-server: cache eviction thread error: Connection timed out [1512735814:255741][22179:0x7f3b7d224700], t, eviction-server: the process must exit and restart: WT_PANIC: WiredTiger library panic [1512735814:255746][22179:0x7f3b7d224700], t, eviction-server: aborting WiredTiger library
The configuration:
############################################ # RUN PARAMETERS ############################################ abort=0 alter=0 auto_throttle=1 backups=0 bitcnt=7 bloom=1 bloom_bit_count=27 bloom_hash_count=29 bloom_oldest=0 cache=37 cache_minimum=20 checkpoints=on checkpoint_log_size=59 checkpoint_wait=94 checksum=uncompressed chunk_size=10 compaction=0 compression=none data_extend=0 data_source=file delete_pct=7 dictionary=0 direct_io=0 encryption=none evict_max=5 file_type=row-store firstfit=0 huffman_key=0 huffman_value=0 independent_thread_rng=1 in_memory=0 insert_pct=12 internal_key_truncation=1 internal_page_max=16 isolation=random key_gap=6 key_max=115 key_min=21 leaf_page_max=17 leak_memory=0 logging=0 logging_archive=0 logging_compression=none logging_file_max=411512 logging_prealloc=1 long_running_txn=0 lsm_worker_threads=3 merge_max=6 mmap=1 modify_pct=63 ops=100000 prefix_compression=1 prefix_compression_min=3 quiet=1 read_pct=5 rebalance=1 repeat_data_pct=38 reverse=0 rows=100000 runs=100 salvage=1 split_pct=82 statistics=0 statistics_server=0 threads=22 timer=360 transaction_timestamps=1 transaction-frequency=82 value_max=926 value_min=19 verify=1 wiredtiger_config= write_pct=13 ############################################
The original failure:
http://build.wiredtiger.com:8080/job/wiredtiger-test-format-stress/59888