On 26 Mar 2015, at 17:18, Scott Ribe <scott_ribe@elevated-dev.com> wrote:
> On Mar 26, 2015, at 10:08 AM, Graeme B. Bell <grb@skogoglandskap.no> wrote:
>>
>> Assuming that you have your server configured with 1 or more hot standbys.
>> Are there situations where it is smart & sane to allow a controlled (slow, not emergency) shutdown to complete
asynchronouslywithout knowing if any standby got the last bits of wal?
>
> Sure.
>
> Point updates. Shut down 9.4.0. Immediately re-start 9.4.1 on the same cluster.
> OS security update, reboot server, PG shuts down and re-starts on boot.
Thanks for the examples, they should have been obvious to me.
You're right, because the master isn't going to throw away wal, it'll be kept due to wal_keep_segments, or 9.4
replicationslots or whatever, and since the master *does* come directly back up in this scenario, 'eventually' will
indeedarrive.
Graeme Bell