database system was not properly shut down

Greetings,

I ran the recent installer from EnterpriseDB
(postgresql-9.3.4-2-windows-x64.exe) on a freshly installed Windows Server
2012 Datacenter with recent patches. Default install, nothing fancy,
next-next-next-finish. So there is now a service "postgresql-x64-9.3" with
automatic startup running commandline "C:\Program
Files\PostgreSQL\9.3\bin\pg_ctl.exe" runservice -N "postgresql-x64-9.3" -D
"C:/Program Files/PostgreSQL/9.3/data" -w".

Everytime I restart the OS I see in pg_log that "database system was not
properly shut down; automatic recovery in progress". Pg_log is not quite
consistent every shutdown, for example:

2014-04-18 15:10:31 CEST LOG: received fast shutdown request
2014-04-18 15:10:31 CEST LOG: aborting any active transactions
2014-04-18 15:10:31 CEST LOG: autovacuum launcher shutting down
2014-04-18 15:10:32 CEST ERROR: canceling statement due to user request

or

2014-04-18 14:36:57 CEST LOG: received fast shutdown request
2014-04-18 14:36:57 CEST LOG: aborting any active transactions
2014-04-18 14:36:57 CEST FATAL: terminating connection due to
administrator command
2014-04-18 14:36:58 CEST FATAL: terminating connection due to
administrator command
2014-04-18 14:36:58 CEST FATAL: terminating connection due to
administrator command
2014-04-18 14:36:58 CEST FATAL: terminating connection due to
administrator command
2014-04-18 14:36:58 CEST FATAL: terminating connection due to
administrator command
2014-04-18 14:36:58 CEST LOG: autovacuum launcher shutting down
2014-04-18 14:36:58 CEST ERROR: canceling statement due to user request
2014-04-18 14:36:58 CEST FATAL: the database system is shutting down
...
2014-04-18 14:37:08 CEST FATAL: the database system is shutting down
2014-04-18 14:37:08 CEST LOG: CreateProcess call failed: A blocking
operation was interrupted by a call to WSACancelBlockingCall.

(error code 19)
2014-04-18 14:37:08 CEST LOG: could not fork new process for connection: A
blocking operation was interrupted by a call to WSACancelBlockingCall.
...
2014-04-18 14:37:09 CEST LOG: CreateProcess call failed: A blocking
operation was interrupted by a call to WSACancelBlockingCall.

(error code 19)
2014-04-18 14:37:09 CEST LOG: could not fork new process for connection: A
blocking operation was interrupted by a call to WSACancelBlockingCall.

or

2014-04-17 17:46:12 CEST LOG: received fast shutdown request
2014-04-17 17:46:12 CEST LOG: aborting any active transactions
2014-04-17 17:46:12 CEST FATAL: terminating connection due to
administrator command
2014-04-17 17:46:12 CEST FATAL: terminating connection due to
administrator command
2014-04-17 17:46:12 CEST FATAL: terminating connection due to
administrator command
2014-04-17 17:46:12 CEST FATAL: terminating connection due to
administrator command
2014-04-17 17:46:12 CEST LOG: autovacuum launcher shutting down
2014-04-17 17:46:12 CEST ERROR: canceling statement due to user request
2014-04-17 17:46:14 CEST FATAL: the database system is shutting down
...
2014-04-17 17:46:23 CEST FATAL: the database system is shutting down
2014-04-17 17:46:23 CEST LOG: server process (PID 1388) was terminated by
exception 0xC0000189
2014-04-17 17:46:23 CEST HINT: See C include file "ntstatus.h" for a
description of the hexadecimal value.
2014-04-17 17:46:23 CEST LOG: terminating any other active server processes
2014-04-17 17:46:23 CEST WARNING: terminating connection because of crash
of another server process
2014-04-17 17:46:23 CEST DETAIL: The postmaster has commanded this server
process to roll back the current transaction and exit, because another
server process exited abnormally and possibly corrupted shared memory.
2014-04-17 17:46:23 CEST HINT: In a moment you should be able to reconnect
to the database and repeat your command.
2014-04-17 17:46:23 CEST LOG: abnormal database system shutdown

The "exception 0xC0000189" I found means "A write operation was attempted
to a volume after it was dismounted." but other than that I'm
hitting dead-ends.

Any suggestions how to troubleshoot this? While the autorecovery is
always successful so far, I don't want to push my luck with a productive
database.

Kind regards,

//Eric