I recently came across intresting issue. I have a sharded cluster which has two replication sets with 2 secondaries. On my primaries, there were 300-400 connections at a point in time, shortly after sometime around a minutes spane, there were almost read and write connections in total 1200 spiked over primaries. Read and write ticket availability went down to 0 for few seconds. Cluster was non responsive during that time. Cluster usually handels more load then this during month end. Any suggestions to find cause of this issue?