On Mon, Jul 31, 2017 at 11:48 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
>> On 7/21/17 06:18, gorka.gil@openbravo.com wrote:
>>> In "Linux" section, show an example of setting shmmax to 16GB, when this
>>> change is not needed from pg 9.3.
>>>
>>> Maybe just remove to avoid confusion?
>>> Maybe also shmall also not needed ?
>
>> It's still needed in some cases, so I don't see why we can't have an
>> example.
>
> I wonder whether it'd make sense to recast those examples as increasing
> semaphore-related settings, as that's much more likely to be needed these
> days. I had to boost the SEMMNI setting on my workstation just last week.
> (Admittedly, I had something like a dozen postmasters going, but still,
> that was an actual change I had to make.) It'd be particularly valuable
> to document this for Linux, with its opaque "kernel.sem" API for the
> semaphore limits.
Agreed. I ran into this many years ago with multiple clusters on a box
and it is definitely non-obvious.