* Removes the initialClusterTime and initialOperationTime session
options.
* Enables causal consistency by default when using an explicit
session.
* Adds a --retryWrites command line option to the mongo shell for
enabling retryable writes in the mongo shell. The retryWrites
options to SessionOptions is left for convenience with testing.
* Renames setClusterTime() to advanceClusterTime(), and adds a
corresponding advanceOperationTime() method to DriverSession.
* Enables assigning transaction numbers for write commands where
ordered=false.
* Prevents the mongo shell from sending afterClusterTime or assigning
transaction numbers when talking to a stand-alone mongod.
* Prevents the mongo shell from assigning transaction numbers when
using an unacknowledged (w=0) writeConcern.
* Changes DBClientRS to re-discover the current primary of the replica
set when it receives an error code representing "not master" in
addition to an error message representing "not master".
* Adds a shellPrint() pretty-printer for SessionOptions and
DriverSession instances so they no longer print out their entire
object definition.
The primary being told to step down will kill all incoming connections
during that stepdown process. When this happens, connections which
are incoming may be closed before they can enter a state which
"reserves" them from closure. To facilitate correct semantics in this
test, we should wait for the primary to completely step down to a
secondary, before trying to authenticate to it again.
There are a few cases where a 3.6 binary will refuse to start up because it
concludes the datafiles belong to a 3.2 mongod. In those cases, the server
must perform a clean shutdown that ensures datafiles are downgraded to a
format that older binaries can understand.
The updates_in_heterogeneous_repl_set.js test wants to call stepUp()
on each of its replica set members to ensure that each one gets an
opportunity to handle inserts and updates. However, we found that the
stepUp()-triggered election was occasionally failing, because even
though a call to awaitReplication() ensured that all members were
caught up on their oplogs, some members were not aware that their
peers were caught up.
The new awaitNodesAgreeOnOpTime() function uses an assert.soon() to
wait until all the members are caught up on their peers' status, so
that the election always succeeds.
