Before embarking on TRT-346 we would like more data on how we're doing with current payloads and aggregation. We know we reject over 60% of payloads, but we do not have data on the reasons for each rejection.

We would like to add the watcher responsibility to try to categorize as many nightly payload failures as possible throughout the week. For now just nightly, but ci would be nice if time allows.

We do not wish to invest in login support in the sippy UI at this time. As such we are proposing a script in sippy repo that will manually be run and update the database.

Rough ideas for a first cut, all open for debate:

• update database table 'release_tags' to include a column for reason for rejection, prefer human readable to ints.
• db password as an env var (PGPASSWORD is default, may as well use that)
• script should be able to list recent payloads that are not categorized for a given stream
• rejected-payloads list 4.11 nightly
• script should be able to categorize with a fixed set of reasons:
• rejected-payloads categorize 4.11.0-0.nightly-2022-06-25-081133
• prompt to select a reason: TEST_FLAKE, CLOUD_INFRA (aws outage or shortage of instance types), RH_INFRA (ci cluster, quay), PRODUCT_REGRESSION, TEST_REGRESSION, text descriptions of each should be displayed.

A curses interface to browse payloads and pick one, then select the reason would be amazing, but may be too much for a first draft.