Re: Using PostgreSQL for service discovery and health-check

On 2/9/23 09:40, Dominique Devienne wrote:
> On Thu, Feb 9, 2023 at 5:51 PM Adrian Klaver <adrian(dot)klaver(at)aklaver(dot)com
> <mailto:adrian(dot)klaver(at)aklaver(dot)com>> wrote:
>
> On 2/9/23 08:16, Dominique Devienne wrote:
> > On Thu, Feb 9, 2023 at 5:05 PM Adrian Klaver
> <adrian(dot)klaver(at)aklaver(dot)com <mailto:adrian(dot)klaver(at)aklaver(dot)com>
> The flip side of that is that with known ports it would it easier to
> have a process on the Postgres machine or in the database that checks
> the ports on regular basis. And as part of that process mark any non
> responding ports as inactive. That would solve the zombie problem.
>
>
> That's one possibility. But the "reaper" process could just as well scan
> the service table,
> and probe those too. So again, I'm not sure what the fixed-port approach
> gains me, beside
> perhaps the reaper not having to connect to PostgreSQL itself. I'm OK
> with connecting.

As to fixed port and pulling vs services pushing, there is a security
side. Not sure who controls the external services, but there is the
chance that someone knowing they exist could inject their own version of
a service/server. With random ports that makes that easier as you would
not know what is canonical. With the pull process you have a
verified(presumably) list of servers and ports they listen on.

>
> Thanks for the your input. Always good to have one's arguments
> challenged by experts.

--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com