[HDDS-5771] Speed up TestDatanodeHddsVolumeFailureToleration by reducing dead interval

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.2.0
• Fix Version/s: 1.2.0
• Component/s: test
• Labels: pull-request-available
• Target Version/s: 1.2.0
• Assignee: Stephen O'Donnell
• Reporter: Stephen O'Donnell
• Created: 21/Sep/21 10:07
• Updated: 21/Sep/21 12:24
• Resolved: 21/Sep/21 12:24

Description

In TestDatanodeHddsVolumeFailureToleration, the tests wait for a long time on a DN to be marked dead by SCM. This is because the stale, dead and heartbeat intervals have not been set low enough, as they are using the default. Changing the various timers, the runtime is reduced from about 4mins to 53 seconds on my laptop.

Further, there are two tests - one that fails one volume on a DN and ensures it can restart ok. Then a test which fails 2 volumes and ensure it fails to start. We can combine these two tests into 1 and save creating a second mini-Cluster. This further reduces the runtime to 37s.

On github, this test runs in:

[INFO] Tests run: 2, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 236.534 s - in org.apache.hadoop.ozone.dn.volume.TestDatanodeHddsVolumeFailureToleration

Issue Links

• links to: GitHub Pull Request #2668