Presentation is loading. Please wait.

Presentation is loading. Please wait.

Testing over Long Duration POD

Similar presentations


Presentation on theme: "Testing over Long Duration POD"— Presentation transcript:

1 Testing over Long Duration POD
Yang (Gabriel) Yu

2 Status Summary Questions from TSC Feedback from Infra Group Next Steps
Stability test should focus on stable release. Is XCI stable for the test?  Test OPNFV installers? How? Round robbin? Feedback from Infra Group Intel-18 could be used for long duration test Connected to sandbox server, could test first before committing to production Next Steps Testing plan on Intel-18 Automated triggered in CI

3 Testing Arrangement (Some Suggestions)
Arrangement 1: Different slots for different projects over OSA Bottlenecks Storperf VSperf Yardstick OSA Mon. Tue. Wen Thu Fri Sat Sun. The time slot assigned to each project depends on its test cases requirement Maybe 1 more installer, however the schedule will be too full to get useful results Weekends are also used which may not be feasible? May need access to the POD to locate the failure Feed back bugs that only exit in long duration tests

4 Testing Arrangement (Some Suggestions)
Arrangement 2: Shared time slot over sets of installers (Recommended) Bottlenecks & Yardstick Storperf & VSperf OSA 2 Weeks for each Time Slot OSA 2 project share a timeslot for join testing, i.e., either run test cases simultaneously (Storage and CPU testing in parallel, extra scripts needed) Or one project call the other to do some tests (in this case, could Bottlenecks act as load manager for Storperf or VSperf?) May need access to the POD to locate the failure Feed back bugs that only exit in long duration tests

5 Testing Arrangement (Some Suggestions)
Arrangement 3: Hybrid time slots usage over sets of installers Bottlenecks & Yardstick OSA Mon Tue Wen Thu Fri Sat. Sun. Storperf VSperf OSA Set of installers Splited time slots for Storperf and VSperf, May not be able to test different installers Maybe 1 more installer besides OSA Set of Installers depends on each time slot (based on projects’ choices) May need access to the POD to locate the failure Feed back bugs that only exit in long duration tests

6 Questions Q1: How the sandbox environment operate? Is it triggered by CI? Q2: Do we need granted access to the POD? If needed, name list? Q3: Do we need a separate directory in Releng to trigger the test? Who is responsible? Q4: Bottlenecks support installation of Prometheus+Collectd+Node. Should Bottlenecks provide test monitoring or just find out if Barometer could help and Bottlenecks contribute in Barometer regarding Prometheus? Q5: Is there need for Bottlenecks to act as load manager to iteratively calling Storperf or VSperf test cases?

7 Thank You


Download ppt "Testing over Long Duration POD"

Similar presentations


Ads by Google