On Thu, 26 Nov 2015 at 22:16 -0700, Bryan Davis wrote:
I made a tool [4] that reads the heartbeat_p database on from the server that hosts each shard and matches it with the shard for each wiki. The tool gets all (dbname, slice) pairs from meta_p.wiki and the slice replag from heartbeat_p.heartbeat from the server hosting each slice and then matching them up in the table. I think I got the logic here right, but you can view the source [5] to see if you agree.
Bryan
Can you add a new set of three column to see the lag per slice to get a more concise way to see lag per slice.