PuppetDB running slowly, nodes not updating
We are having trouble with the PE console saying all our nodes are unresponsive, and PuppetDB is being hammered by the following query:
DELETE FROM edges WHERE certname=$1 and trim(leading '\x' from source::text)=$2 and trim(leading '\x' from target::text)=$3 and type=$4
Notes: We are running PE 2015.2.2 (about to update) in a split configuration. All three servers agree on the time, and are using the same time source. Nodes are running puppet agent -t successfully. Catalog and fact updates are appearing in the logs on the master.
Does anyone know what this is?