Grouping issues here:
- mediacounts job non-deterministic problem - jobs fail reguarly, rerun succeed most of the time https://hue.wikimedia.org/oozie/list_oozie_coordinator/0002893-210211141618764-oozie-oozi-C/
- mediarequest hourly job used a CTE with a UDF that output a struct [partial patch, see comment below]
- projectview goe job used reserved keyword "rows"
- interlanguage job syntax problem
- all cassandra jobs failed due to logging config conflict, all restarted now except per_file
- mediarequests per_file broken due to keywords timestamp and user, but union all worked
- event refine not working due to spark being bundled with hadoop, Andrew has a temporary workaround
- clickstream job problem with SparkMain - https://gerrit.wikimedia.org/r/c/operations/puppet/+/663582/
- apis job killed, same SparkMain problem
- data quality coordinators failing due to a Hive/SQL exception
- geoeditors monthly job is failing due to use of reserved word 'date'
- T274384: Repackage spark without hadoop, use provided hadoop jars
- Virtualpageview hourly error: invalid function ''project'', see T274322#6828807 (deferred due to probable migration to Spark SQL)
- Webrequest table sample see T274322#6828810 (deferred due to probable migration to Spark SQL)
- Nullpointer for pageview_actor hourly see T274322#6828818 (deferred due to probable migration to Spark SQL)
- Lots of reportupdater jobs using reserved keywords (mostly "date", but one file using "timestamp")
- Pageviews per-country AQS job change errors
- T277822