From: Michael Paquier [mailto:michael@paquier.xyz]
> Thanks for the new version. Per my comments up-thread here, you cannot
> actually use PGC_BACKEND:
> https://www.postgresql.org/message-id/20180919061303.GB19808@paquier.x
> yz
>
> This would break the case where this parameter is reloaded when a session
> does not use a custom value for wal_sender_timeout. I have also looked
> at all the code paths using wal_sender_timeout, and the change looks safe
> to me. Please find attached an update, simplified, version.
> Does that look fine to you?
Thank you, PGC_USERSET is off course fine to me. I thought PGC_BACKEND includes the capability of PGC_SIGHUP, because
PGC_BACKENDis placed after PGC_SIGHUP in the definition of enum GucContext. That was a pitfall. I should have paid
moreattention.
I find it more user friendly to include a description somewhere that the user can tune the timeout per standby, like I
addeda tip in the description of wal_sender_timeout. I'm afraid users won't know whether and how to tune the setting
perstandby, as libpq's options parameter doesn't seem well-known in my experience.
Regards
Takayuki Tsunakawa