"[conn557392] killcursors: found 0 of 1" in the Mongodb replication main log

Question

Now I am launching production mongodb replication with version 2.6.

Today I found out that the main instance of mongod continues to write the log:

[conn557392] killcursors: found 0 of 1

And I checked db.serverStatus (). metrics.cursor, there is actually a large amount of time cursor, as mentioned in this discussion .

My questions

• Since I set all of my reading logic to be preferred Secondary, the primary should be write-only. Why does he need to kill the cursor? Does the cursor assume it's read-only?

• Why aren't all application services affected, although there are more than half a million paced timedOut? Am I taking this seriously?

Thanks in advance!