Re: a vulnerability in PostgreSQL - Mailing list pgsql-hackers

Or else people in our situation where it takes forever to upgrade the
software because of its heavy use and the risk involved in upgrading, not
to mention the problems encountered when we did test-runs of the upgrade.

Then there is always the thorny issue of loads of software that uses the
databases, most of it under user control and any incompatibility between
versions,
no matter how small, could have horrible implications for our clients and
therefore, us.

You see, we are an ISP and a consultancy specialising in database-driven
web sites and corporate infrastructure. We based nearly everything that
we
do on PostgreSQL and although we upgrade when we can, our hands are
very tied. For us, patching is a necessity, not an option. To migrate
a client means rebuilding an entire UAT (user acceptance test site) for
them, extensively testing what we can ourselves, then asking the
client to allocate time, money and people to test their systems and their
own code as well. They will also need to allocate development time, money
and people to fix any problems that they find in compatibility.

Then there is the throny issue of both companies needing to synchronise
their resources and schedules so that we can work together solving any
problems that arise.

Finally, because we have no control over the customer's quality control,
and
because customers very often don't have the inhouse expertise to even
understand
what a proper test is about (they most often have hired in expensive
consultants
or contracted other companies to do the work), we have no guarantees that
when the
client thinks that they have tested the site, they really have. Now most
people
will say "that's their problem", but you see, it isn't. Because these are
business-critical systems that we're talking about and the change was
initiated
by us. So if the system fails, for whatever reason, on the new software,
even
if it isn't anything to do with the new software, we get the flack. And
also in
the clients' eyes, we are responsible because their system "was working
fine until
[we] forced and upgrade to new software".

And that is customer relations being damanged, even if the customer is
wrong.

See the problem? I am only writing this to add to the pool of knowledge
about
how PG is used and the real-world implications of PG being used for
business
critical or customer-image systems. PG is extremely well suited for this
and
the benefits over closed-source systems is enormous, not to mention the
fact
that PG has email lists like this one with all the fine brains on this
list
pooled together. It shows in the code and it shows in the satisfaction
level
of those who use it (I have never once had a client who was dissatisfied
with
PG. Most clients were surprised that OpenSource existed, that it is free
and
that it is such great quality without any catches or got-yous that
normally
comes with "free" things from commercial companies.).

Well, that's my hat in the ring. Hope that it helps someone out there or
at
least adds something to our pooled knowledge!

Brad
Kieser.net


>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<

On 5/3/02, 4:43:31 AM, Lincoln Yeoh <lyeoh@pop.jaring.my> wrote regarding
Re: [HACKERS] a vulnerability in PostgreSQL :


> I hope you won't make this standard practice. Because there are quite
> significant differences that make upgrading from 7.1.x to 7.2
troublesome.
> I can't name them offhand but they've appeared on the list from time to
time.

> For 6.5.x to 7.1.x I believe there are smaller differences, even so there
> might be people who would patch for security/bug issues but not upgrade.
> I'm still on Windows 95 for instance (Microsoft has stopped supporting it
> tho :( ). I think there are still lots of people on Oracle 7.

> Yes support of older software is a pain. But the silver lining is: it's
> open source they can feasibly patch it themselves if they are really hard
> pressed. If the bug report is descriptive enough DIY might not be so bad.
> And just think of it as people really liking your work :).

> Any idea which versions of Postgresql have been bundled with O/S CDs?

> Regards,
> Link.

> At 10:23 AM 5/2/02 -0400, Tom Lane wrote:
> >Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> > > Here are the precise conditions to trigger the scenario:
> >
> > > (1) the backend is PostgreSQL 6.5.x
> > > (2) multibyte support is enabled (--enable-multibyte)
> > > (3) the database encoding is SQL_ASCII (other encodings are not
> > >     affected by the bug).
> > > (4) the client encoding is set to other than SQL_ASCII
> >
> > > I think I am responsible for this since I originally wrote the
> > > code. Sorry for this. I'm going to make back port patches to fix the
> > > problem for pre 7.2 versions.
> >
> >It doesn't really seem worth the trouble to make patches for 6.5.x.
> >If someone hasn't upgraded yet, they aren't likely to install patches
> >either.  (ISTR there are other known security risks in 6.5, anyway.)
> >If the problem is fixed in 7.0 and later, why not just tell people to
> >upgrade?
> >
> >                         regards, tom lane
> >
> >---------------------------(end of broadcast)---------------------------
> >TIP 4: Don't 'kill -9' the postmaster



> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
>     (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)