Recent discussions have revolved around understanding why job run times have increased, when new tests were added, how long individual tests run, how frequently individual tests fail as well as how often individual jobs fail have led to us wanting to summarize this information on a daily basis for reporting / investigation across releases without having to produce large queries each time.

Our jobs and junit table contain this data already, we should create new scheduled queries to summarize based on

Jobs

• min,max,avg,p95 run times
• min,max,avg test counts
• total runs, run failures, run passes (could be calculated)
• job name

Should we also track Job step timings?  Need to research what we capture but we could potentially get rough timing s for getting leases, install, upgrade, conformance, must-gather, gather-extra, etc.

For junits

• min,max,avg,p95 run times
• total executions, execution failures, run passes(could be calculated)
• release
• variants (platform, suite, featureset, topology, arch, owner, etc?)

For starters we can keep the junit summary coarse as we can always go back to the junit table for more details queries for shorter time periods.

The goal for these new table is to be able to provide comparisons between releases / different time periods to analyze overall health and changes within jobs / tests.  Additionally data from these tables can be used to source work like TRT-2375