MongoDB local.oplog.rs query running for long time

Question

I have 2 node cluster(8 vCPUs, 52 GB) for mongodb(3.2.0). I was debugging couple of queries that took several seconds, I ran db.currentOp() to see what queries are running and how much time they are taking and I found a query regarding local.oplog.rs that was running for more than 4000 seconds, following is the query in question

         {
            "desc" : "WT RecordStoreThread: local.oplog.rs",
            "threadId" : "139625785952000",
            "active" : true,
            "opid" : "rs0:145981225",
            "secs_running" : 4168,
            "microsecs_running" : NumberLong("4168129663"),
            "op" : "none",
            "ns" : "local.oplog.rs",
            "query" : {

            },
            "numYields" : 0,
            "locks" : {

            },
            "waitingForLock" : false,
            "lockStats" : {
                "Global" : {
                    "acquireCount" : {
                        "r" : NumberLong(1),
                        "w" : NumberLong(1)
                    }
                },
                "Database" : {
                    "acquireCount" : {
                        "w" : NumberLong(1)
                    }
                },
                "oplog" : {
                    "acquireCount" : {
                        "w" : NumberLong(1)
                    }
                }
            }
        },

Could any please shed some light on this? what is this query doing and should I be concerned?

Answer 1

These are threads related to replication. They're part of normal operation.

• WT RecordStoreThread: local.oplog.rs
• rsSync
• rsBackgroundSync
• ReplBatcher

There's actually an open bug about showing these results by default when they should probably be hidden because they're always long running operations: https://jira.mongodb.org/browse/SERVER-28257

From the bug report:

At present db.currentOp() by default shows replication-related threads such as:

"WT RecordStoreThread: local.oplog.rs"

"rsSync"

"rsBackgroundSync"

"ReplBatcher"

Often these operations are long running and users may get a wrong impression that there is a performance problem.