Cluster stopped replicating

[kripper]

Description

A 2-node CouchDB cluster stopped replicated changes from node-2 to node-1.
Changes were written to node-2.
After node 2 was restarted, replication to node-1 recovered.
Both nodes are on different datacenters (90ms away).

Is this a known issue under certain circumstances?
How can I debug this issue?
Log contains passwords and I didn't see any suspecious error.
What should I look for?

Your Environment

Same setup on both nodes:

• CouchDB version used: couchdb-3.1.1-1.el7.x86_64
• Operating system and version: CentOS 7 - 3.10.0-327.el7.x86_64