Re: bootstrap pg_shseclabel in relcache initialization

Adam Brightwell <adam(dot)brightwell(at)crunchydata(dot)com> writes:
> On Tue, Nov 10, 2015 at 9:18 AM, Adam Brightwell
> <adam(dot)brightwell(at)crunchydata(dot)com> wrote:
>>> I'm with Alvaro: the most interesting question here is why that mistake
>>> did not blow up on you immediately. I thought we had enough safeguards
>>> in place to catch this type of error.

>> Ok, I'll explore that a bit further as I was able to build and use
>> with my hook without any issue. :-/

> Ok, I have verified that it does indeed eventually raise a warning
> about this. It took a few for it to occur/appear in my logs. I was
> expecting it to be more "immediate".

Hm. Salesforce had also run into the issue that the warnings are just
warnings and relatively easy to miss. What we did there was to change
the elog(WARNING)s near the bottom of load_relcache_init_file to
elog(ERROR), but I'm not sure if I want to recommend doing that in the
community sources. In commit 5d1ff6bd559ea8df I'd expected that the
WARNINGs would certainly show up in regression test output, and I thought
I'd verified that that was the case --- did that not happen for you?

regards, tom lane