TailableIterator can time out if kept alive for longer than MongoDB cursor timeout

[laserval]

It seems that a TailableIterator that is kept alive for longer than the default MongoDB cursor timeout of 10 minutes, will be garbage collected in the MongoDB server. See https://jira.mongodb.org/browse/SERVER-12200

One case this will happen is if some service is listening to processed, failed or discarded documents in the oldDocuments collection.

Some solutions:

• Set the MongoDB option NOTIMEOUT on the tailable cursor
• Recreate the cursor if it fails
• Await fixes to timeouts in MongoDB and mongo-java-driver and require newer versions of both (https://jira.mongodb.org/browse/SERVER-8188, https://jira.mongodb.org/browse/SERVER-12200)

Related issue concerning tailable and awaitdata: https://jira.mongodb.org/browse/JAVA-1091