The long tests take almost 9 hours to run.
The following tests are required to run on the directory (WT_TEST) generated by the execution of the 500m-btree-populate test.
- 500m-btree-50r50u
- 500m-btree-50r50u-backup
- 500m-btree-80r20u
- 500m-btree-rdonly
At this stage, we are simply running all these tests under a single task sequentially and uploading the result generated from each task to evergreen and Atlas.
- func: "run-perf-test" vars: perf-test-name: 500m-btree-populate maxruns: 1 wtarg: -args ['"-C create,statistics=(fast),statistics_log=(json,wait=1,sources=[file:])"'] -ops ['"load", "warnings", "max_latency_insert"'] - func: "upload-perf-test-stats" vars: perf-test-name: 500m-btree-populate - func: "run-perf-test" vars: perf-test-name: 500m-btree-50r50u maxruns: 1 wtarg: -args ['"-C create,statistics=(fast),statistics_log=(json,wait=1,sources=[file:])"'] -ops ['"read", "update", "warnings", "max_latency_read_update"'] - func: "upload-perf-test-stats" vars: perf-test-name: 500m-btree-50r50u ...
However, running the tests in parallel should save time.
Scope:
- Upload the directory generated from the 500m-btree-populate test to an s3 bucket.
- Run the tests mentioned above in separate tasks. Each task will
- Fetch the directory from the S3 bucket
- Run the test on the directory.
- related to
-
WT-5014 Migrate jenkins wiredtiger perf long job to evergreen
- Closed