The purpose of suppress_ragged_eofs=False was to micro-optimize the
normal case: when there's no EOF.

But commit 061cd5d8 showed that this
option only turns ragged EOF into an exception. It may be easier for
alternate NEO implementations to close the SSL connection properly. Or
the performance benefit was not worth the risk to freeze a NEO process.

Testing NEO/go client wrt NEO/py server revealed a bug in NEO/py SSL
handling: proper non-ragged EOF from a peer is ignored, and so leads to
hang in infinite loop inside _SSL.receive with read_buf memory growing
indefinitely. Details are below:

NEO/py wraps raw sockets with

	ssl.wrap_socket(suppress_ragged_eofs=False)

which instructs SSL layer to convert unexpected EOF when receiving a TLS
record into SSLEOFError exception. However when remote peer properly
closes its side of the connection, socket.read() still returns b'' to
report non-ragged regular EOF:

https://github.com/python/cpython/blob/v2.7.18/Lib/ssl.py#L630-L650

The code was handling SSLEOFError but not b'' return from socket recv.
Thus after NEO/go client was disconnecting and properly closing its side
of the connection, the code started to loop indefinitely in _SSL.receive
under `while 1` with  b'' returned by self.socket.recv() appended to
read_buf again and again.

-> Fix it by detecting non-ragged EOF as well and, similarly to how
SSLEOFError is handled, converting them into self._error('recv', None).

See merge request nexedi/neoppod!17

The scenario that was described in comments was meaningless
because S1 never goes out-of-date.

For the master, the purpose of -m/--masters is to specify addresses
of other master nodes, since its own address is already known via
-b/--bind. Therefore, an empty value for -m/--masters is valid.
The user remains free to repeat the -b value in -m.

More generally, a node may choose to only specify master addresses
via -D/--dynamic-master-list, so the check that at least one master
address is specified is moved where the NodeManager is expected to be
initialized.

The time complexity of previous one was too bad. With several tens of
concurrent transactions, we saw commits take minutes to complete and
the whole application looked frozen.

This new algorithm is much simpler. Instead of asking the oldest
transaction to somewhat restart (we used the "rebase" term because
the concept was similar to what git-rebase does), the storage gives
it priority and the newest is asked to relock (this request is ignored
if vote already happened, which means there was actually no deadlock).

testLocklessWriteDuringConflictResolution was initially more complex
because Transaction.written (client) ignored KeyError (which is not the
case anymore since commit 8ef1ddba).

This is all the more important for RocksDB that it wants to keep all
transaction work in RAM.

Once we had to truncate 40% of a 1TB MyRocks DB with 24 partitions,
4 being processed in parallel. Even when committing between partitions,
MariaDB used up to 200 GB. Without the commit, 1TB RAM would not have
been enough.

The default value is quickly exceeded when truncating a DB.
Obviously, you may need a lot of RAM.

Resetting a storage node could mark all TEST log entries as being
emitted by this storage node. For example:

16:18:12.9114 S2         #0x0007 AskStoreObject                 > S1 (...)

======================================================================
FAIL: check_tid_ordering_w_commit (neo.tests.zodb.testBasic.BasicTests)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "ZODB/tests/BasicStorage.py", line 397, in check_tid_ordering_w_commit
    self.assertEqual(results.pop('lastTransaction'), tids[1])
  File "neo/tests/__init__.py", line 301, in assertEqual
    return super(NeoTestBase, self).assertEqual(first, second, msg=msg)
failureException: '\x03\xd8\x85H\xbffp\xbb' != '\x03\xd8\x85H\xbfs\x0b\xdd'

This requires ZODB >= 5.6.0

This fixes the bug that with only email notification, monitoring
stopped checking whether backup clusters are lagging after status is
unchanged since the last check (about lagging, what is compared is
the set of lagging backups). Until another event wakes up monitoring.

The code is also simplified in that there's no need for the moment to
have a different timeout between the normal case and a smtp failure.

When concurrent transactions fail with different storages (e.g. only network
issues between C1-S2 and C2-S1), in such a way that each transaction can be
committed but not both (or the cluster would be non-operational), and if the
first transaction is aborted (between tpc_vote and tpc_finish), then the second
wrongly failed with INCOMPLETE_TRANSACTION.

And if both transactions could be committed (e.g. more than 1 replica),
some nodes would be disconnected for nothing.

This fixes:

  Traceback (most recent call last):
    ...
    File "neo/admin/handler.py", line 200, in answerLastTransaction
      app.maybeNotify(name)
    File "neo/admin/app.py", line 380, in maybeNotify
      self._notify(False)
    File "neo/admin/app.py", line 302, in _notify
      body += '', name, '    ' + backup.formatSummary(upstream)[1]
    File "neo/admin/app.py", line 74, in formatSummary
      tid = self.backup_tid if backup else self.ltid
  AttributeError: 'Backup' object has no attribute 'backup_tid'

This fixes:

  Traceback (most recent call last):
    File "neo/master/app.py", line 172, in run
      self._run()
    File "neo/master/app.py", line 182, in _run
      self.playPrimaryRole()
    File "neo/master/app.py", line 314, in playPrimaryRole
      self.backup_app.provideService())
    File "neo/master/backup_app.py", line 101, in provideService
      app.changeClusterState(ClusterStates.STARTING_BACKUP)
    File "neo/master/app.py", line 474, in changeClusterState
      ) or not node.isClient(), (state, node)
  AssertionError: (<EnumItem STARTING_BACKUP (4)>, <ClientNode(uuid=C1, state=RUNNING, connection=<ServerConnection(nid=C1, address=127.0.0.1:52430, handler=ClientReadOnlyServiceHandler, fd=59, on_close=onConnectionClosed, server) at 7f38f5628390>) at 7f38f5628ad0>)

In such case, it didn't reconnect, but thought it was connected,
which eventually led to crashes like:

  Traceback (most recent call last):
    ...
    File "neo/admin/handler.py", line 130, in answerClusterState
      self.app.updateMonitorInformation(None, cluster_state=state)
    File "neo/admin/app.py", line 274, in updateMonitorInformation
      self.upstream_admin_conn.send(Packets.NotifyMonitorInformation(kw))
    File "neo/lib/connection.py", line 565, in send
      raise ConnectionClosed
  neo.lib.connection.ConnectionClosed

This fixes:

  Traceback (most recent call last):
    File "neo/scripts/neoadmin.py", line 31, in main
      app.run()
    File "neo/admin/app.py", line 179, in run
      self._run()
    File "neo/admin/app.py", line 199, in _run
      self.em.poll(1)
    File "neo/lib/event.py", line 155, in poll
      self._poll(blocking)
    File "neo/lib/event.py", line 220, in _poll
      if conn.readable():
    File "neo/lib/connection.py", line 487, in readable
      self._closure()
    File "neo/lib/connection.py", line 545, in _closure
      self.close()
    File "neo/lib/connection.py", line 534, in close
      handler.connectionFailed(self)
    File "neo/admin/handler.py", line 210, in connectionClosed
      app.connectToUpstreamAdmin()
    File "neo/admin/app.py", line 230, in connectToUpstreamAdmin
      None, None, self.name, None, {}))
    File "neo/lib/connection.py", line 574, in ask
      raise ConnectionClosed
  neo.lib.connection.ConnectionClosed