https://sippy.dptools.openshift.org/sippy-ng/tests/4.14/analysis?test=Cluster%20upgrade.%5Bsig-cluster-lifecycle%5D%20cluster%20upgrade%20should%20complete%20in%2075.00%20minutes

This test fails most of the time. Aggregation succeeds since this failure rate is baked in, but it'd be nice to fix the test.

If you look at the analysis above, the nurps affected by this aren't many.  The 75 minute threshold only applies to SDN jobs because we always add 15 for OVN, and we always add an additional 30 for AWS.

On blocking jobs, the only sdn upgrades we have are aws micro and azure minor.

Our 30-day average for azure sdn minor are 69 minutes:

sippy_openshift=> select avg(pjt.duration) from prow_job_run_tests pjt inner join prow_job_runs pjr on pjr.id = pjt.prow_job_run_id inner join prow_jobs pj on pj.id = pjr.prow_job_id where 'azure' = ANY(variants) AND 'sdn' = ANY(variants) AND 'upgrade-minor' = ANY(variants) and pj.release = '4.14' and pjt.test_id = 44 and pjr.timestamp > current_date - interval '30' day;

        avg        
-------------------
 4176.410003778753
 

We should probably bump the limit, and also maybe figure out a better way to handle upgrade duration measurements.