some hopefully heavy mogilefs development

[Mark Smith]

> My company is diving head first into some large scale file hosting and
> we think mogilefs will serve our needs well.  Today we got four pretty
> nice boxes and have a cluster of another say twenty not as nice boxes
> that we can use to pound on our mogile cluster.  Had we known about the
> summit last week (sorry, new to the list), we would have likely tried
> pretty hard to make it.

Glad to hear it - sorry the timing didn't work out!

> We are going to have two and a half programmers on the project for a
> fair amount of time and would like to do things right and give back what
> we can.  Keeping in mind that we are going to have our own API in front
> of mogile, we would like to do at least a couple things:
> 
> 1.       Write a test suite (this would likely test through our API, but
> perhaps we could write start some mogile specific tests.  I am not sure
> what is out there already)

We have the beginnings of a MogileFS test suite - adding on to that is
going to be 100% welcome.  It certainly needs a lot more tests.

> 2.       Write a hammer suite.  Just hammer the heck out of mogile and
> see how it holds up

This is one thing people mentioned wanting.  I know we'd want it too,
as we're constantly wanting to test changes in Mogile that may impact
performance, etc.

> Just wondering if anyone has much experience with either.  Also
> wondering if we should hold off for mogilefs2.0, though the slides did
> say things would be backwards compatible.

The SVN trunk is already on 2.0.  It's stable, we're using it in production
on LiveJournal.com.  Since you're starting a new project, I'd recommend
starting with MogileFS 2.0 if you're willing to help us test it and stress
it and look for any issues.


--
Mark Smith
junior at danga.com