Re: [GENERAL] PostgreSQL's stability,

On Sun, 23 Aug 1998, Rostislav Matl wrote:

>
> Hi,
> Browsing through MySQL pages I've found crashme tests including
> note about PostgreSQL that "Anybody with access to database
> can get it down". It's that true ? I've not mentioned such
> behaviour...

There have been, in the past, several bugs that resulted in
corruption of the shared memory error that all postgres processes use.
The result of that corruption was that all the backends would either die
out, or require the superuser to kill them off and restart them.

As we have 'evolved', and those bugs are reported, they have been
addressed, and the likelihood of that happening in current releases has
dropped dramatically.

The more ppl that have access to the server, the better the chance
of it happening, since you have better odds that someone will happen to
trigger it.

If this is a large concern, though, what you can do is setup a
'server' for each client/user wishing to use it, so that they have their
own virtual area to work in, so that one person doesn't affect another.
It uses up more resources that way, but if you are more concerned with one
client not affecting another, and not with resources, then that would be a
safer way to go...

Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy(at)hub(dot)org secondary: scrappy(at){freebsd|postgresql}.org