-
Type: Bug
-
Resolution: Unresolved
-
Priority: Minor - P4
-
None
-
Affects Version/s: None
-
Component/s: Test wtperf
-
Storage Engines
-
StorEng - Defined Pipeline
When wtperf completes its workload operations, it prints out a summary that looks like this:
0 inserts, 0 modifies, 59532 reads, 0 truncates, 3576 updates, 0 backups, 0 checkpoints, 0 flush_tiers, 0 scans in 5 secs (640 total secs) Executed 0 insert operations (0%) 0 ops/sec Executed 0 modify operations (0%) 0 ops/sec Executed 2663415 read operations (86%) 8878 ops/sec Executed 0 truncate operations (0%) 0 ops/sec Executed 427757 update operations (13%) 1425 ops/sec Executed 1 backup operations Executed 15 checkpoint operations Executed 0 flush_tier operations Executed 0 scan operations
The bug is that all of the system operations such as backup and checkpoint may still be in progress. In particular, I've seen this issue with backups that take a long time. The program correctly waits for all the system threads to complete and exit, but backup 2 was in progress when the workload completed and the program is incorrectly reporting only 1 backup executed.
wtperf should wait until all the system threads complete before reporting the non-workload information, at a minimum.