MongoDB - 100% CPU - recovering shard (after killing updates)

Can a recovering shard (as shown below) drive CPU to 100% for hours and hours? I have killed all the queries and updates (based on db.currentOp()) and stopped the flow of new data, but the CPU stays at 100%. I can run little queries, but deleting 200 rows (via an index) runs for hours. Inserts are getting backed up and not performing timely. We have a large collection (almost 200 GB) with about 5 indexes. Can the server be doing internal clean-up that runs the CPU so high that our normal flow of data bogs down? (running cloud/Atlas) In the db.currentOps(), I see quite a few that look like the json below; these don't seem to represent any queries, inserts, updates, or deletes that we are running or started. { "type": "op", "host": "atlas-10ydpn-shard-00-04.otwxp.mongodb.net:27017", "desc": "conn1280774", "connectionId": 1280774, "client": "54.221.39.10:59692", "clientMetadata": { "driver": { "name": "PyMongo", "version": "3.11.0" }, "os": { "type": "Linux", "name": "Linux", "architecture": "x86_64", "version": "4.14.193-110.317.amzn2.x86_64" }, "platform": "CPython 3.7.9.final.0" }, "active": true, "currentOpTime": "2020-10-22T12:52:12.100+0000", "effectiveUsers": [ { "user": "pythonapp", "db": "admin" } ], "opid": 20478470, "lsid": { "id": "059e860217054269a3f30c93a7b9c9cb", "uid": null }, "secs_running": 2370, "microsecs_running": 2370403369, "op": "command", "ns": "relatixtd.$cmd", "command": { "update": "ProdPolygonIOQuotes", "ordered": true, "writeConcern": { "w": "majority" }, "lsid": { "id": "059e860217054269a3f30c93a7b9c9cb" }, "txnNumber": 1, "$clusterTime": { "clusterTime": null, "signature": { "hash": null, "keyId": 6857244042040705026 } }, "$db": "relatixtd" }, "numYields": 1, "waitingForLatch": { "timestamp": null, "captureName": "ReplicationCoordinatorImpl::_mutex" }, "locks": {}, "waitingForLock": false, "lockStats": { "ParallelBatchWriterMode": { "acquireCount": { "r": 4 } }, "ReplicationStateTransition": { "acquireCount": { "w": 6 } }, "Global": { "acquireCount": { "r": 2, "w": 3 } }, "Database": { "acquireCount": { "r": 1, "w": 3 } }, "Collection": { "acquireCount": { "r": 1, "w": 3 } }, "Mutex": { "acquireCount": { "r": 4 } } }, "waitingForFlowControl": false, "flowControlStats": { "acquireCount": 2, "timeAcquiringMicros": 1 } },