On 9/26/16 8:53 AM, Tom Lane wrote:
> I think that it's 100% pointless for get_control_dbstate
> to be worried about transient CRC failures. If writes to pg_control
> aren't atomic then we have problems enormously larger than whether
> "pg_ctl promote" throws an error or not.
The new code was designed to handle that, but if we think it's not an
issue, then we can simplify it. I'm on it.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services