High CPU Utilization on Sync Gateway server after receiving "Error processing DCP stream error"

We are running into a frequent issue where our sync gateway info log will show an error listed below and CPU will spike to 90-100%. The syncing process stops right after this error and does not continue until we restart the sync gateway.

This is what I have noticed in the sg_info.log.

*Error processing DCP stream - will attempt to restart/reconnect if appropriate: client.Transmit, err: write tcp> write: broken pipe. – base.(DCPReceiver).OnError() at dcp_feed.go:134
2019-11-02T19:48:38.113Z [INF] Sync: c:[5e1390e9] Sent 200 changes to client, from seq 16818832:16660817

Next noticeable entry:

*2019-11-06T18:26:55.396Z [WRN] DCP RollbackEx request - rolling back DCP feed for: vbucketId: 343, rollbackSeq: 0. – base.(DCPReceiver).RollbackEx() at dcp_feed.go:294

2019-11-06T18:26:56.771Z [INF] DCP: Received #30094389 after 340186633ms

Has anyone seen these errors and what may be causing them?