Re: PITR: Request for assistance with alpha test plan

> -----Original Message-----
> From: pgsql-hackers-pitr-owner@postgresql.org
[mailto:pgsql-hackers-pitr-
> Subject: [pgsql-hackers-pitr] PITR: Request for assistance with alpha
test
> plan
>
>
> A detailed proposal has now been published for PITR functionality.
>
> I'd like to now request assistance from anybody and everybody that
USEs
> PostgreSQL, not just hackers, to work out a well-documented and
> comprehensive test plan for PITR. (Some have offered, though
volunteers
> are required, so I would not make individual requests).
>
> Regression tests currently protect PostgreSQL from a range of ugly
> situations. PITR will need a similarly rigorous approach, though this
> cannot be easily added as regression tests because of the very nature
of
> the PITR functions and possible failure scenarios.

Should be difficult to incorporate it in current regression tests
framework.
We will need a case for backend crash and then startup and recover. Then
we will need a case where we would take backup of a datafile, perform
some insert and updates, corrupt datafile with random data, shutdown
backend, startup, fail to startup, restore datafile from backup, apply
roll forward by applying WAL's and startup. The to test this strategy
thoroughly we will need to pull the plug on running system a couple of
times.


>
> My suggestion is the coordinated assembly of a comprehensive test
plan,
> BY USERS, for USERS. When such a test plan document is assembled, the
> tests described should be able to be repeated many times by many users
> on many platforms, though under a range of workloads, to ensure all
bugs
> are found and reported.
>
> This is ideally suited to those who USE rather than extend the
existing
> functionality. You are skilled people whose contribution is gratefully
> received in this endeavour.
>
> Based upon the possible failures already identified, it should be
> possible to write a set of tests that check for those failure
> conditions. In doing so, it may occur to you that other failure
> conditions may exist also and it would be good to raise those as
> possibilities now.
>
> This ISN'T a call for beta-testers, but a call for assistance in
> assembling the list of tests that alpha-testers will be asked to
> complete. If you don't understand what I'm asking for, then it may not
> yet be time for you to help; however, I am particularly interested in
> assistance from those who know they are using relatively obscure parts
> of the PostgreSQL product.
>
> Based upon volume of response, you may not receive individual
> thanks...but there will be a list of credits somewhere down the
line....

I will be more than ready to participate.

>
> Best Regards, Simon Riggs

Nick