Is this still broken? I can see that new revisions are going in to the revision table when I sql dewiki (or directly to the mentioned hosts with mysql -h labsdb<#>.eqiad.wmnet -u <my username> dewiki_p -p) and run "select max(rev_timestamp) from revision;"

There is no current issue with either dewiki and wikidatawiki, and revisions are coming in with no significant lag (subsecond).

Occasionally, a long-running query that locks tables can delay replication for several hours before it is found and killed, perhaps that is what happened here.

The replication had been stopped some minutes after a wikidatawiki db schema change was executed on wmf cluster to enable wikidata as langlinks repository on wikidata pages. The problem seems to be fixed since the night from Thursday to Friday morning to days laters