Looking at the jobs that were supposed to run over the weekend, one set is stalled waiting for two missing partitions. Looking in hdfs I see input data available in /wmf/data/raw/..., but the refined output locations instead contain _REFINE_FAILED files.
Could these be re-run? Alternatively, if refine will simply not run on these hours, could we place _SUCCESS files in the directories so downstream oozie tasks continue?
$ hdfs dfs -ls /wmf/data/event/mediawiki_cirrussearch_request/datacenter=eqiad/year=2020/month=2/day=*/hour=*/_REFINE_FAILED Picked up JAVA_TOOL_OPTIONS: -Dfile.encoding=UTF-8 -rw-r--r-- 3 analytics analytics 26 2020-02-03 10:21 /wmf/data/event/mediawiki_cirrussearch_request/datacenter=eqiad/year=2020/month=2/day=3/hour=6/_REFINE_FAILED -rw-r--r-- 3 analytics analytics 26 2020-02-03 10:22 /wmf/data/event/mediawiki_cirrussearch_request/datacenter=eqiad/year=2020/month=2/day=3/hour=7/_REFINE_FAILED