1.2.4 now freely available
> Lustre 1.2.4 is only going to be of use
> if they "forward port" it to a current
> kernel, otherwise what would be the use
> of it? It sounds as though it'll be made
> "public" around the time Linux 2.6.11 or
> 2.6.12 will be out. Many users may be
> using an out-of-the-box distribution,
> such as Fedora, and therefore be at
> around Linux 2.6.9. Any public release
> of Lustre will have to work on kernels
> in this range, if there is to be any
> demand for it.
Although we have not yet released new binary packages, we have included our latest supported kernel patches (for Red Hat Enterprise Linux 3's 2.4.21-based kernel, and SuSE Linux Enterprise Server 9's 2.6.5-based kernel) in today's public release of Lustre 1.2.4. We'll make new binary RPMs available shortly, but decided not to release the old ones from July, which are based on vendor kernels with now-known security vulnerabilities. In the meantime, please feel free to compile your own.
We don't currently support kernel.org 2.6 kernels, both because our customers don't run them, and they don't seem very stable (the two are probably quite related). There are people who maintain and occasionally release new Lustre patches for kernel.org kernels, however; you'll find talk of it in the lustre-discuss archives.
> I'm not going to hold my breath - it
> costs $1,500 per site (if I understand
> their pricing scheme correctly) which
> would put it out of the range of anyone
> who just wanted to share the code. (Tech
> support calls are also fairly steep, at
> $300 per hour.)
Just for the record, we typically do not charge by the hour for support. We sell flat-rate annual support agreements, the cost of which is based on the size and complexity of the deployment.
One of our goals in building Lustre is to produce an excellent cluster file system for the Linux community, and so we will continue to make regular releases to the general public. But we also want to continue to improve Lustre and support our customers, which means we need to stay in business. It doesn't do anyone any good if we close up shop, and send our dozens of full-time Lustre developers to find new jobs.
Impatient organizations who need the latest features and fixes, or timely and guaranteed support, can pay for access to the newest releases. Others can wait, and be assured that they will eventually get the code that today's customers are paying for. There's no sinister agenda, no attempt to "take Lustre proprietary", just a balancing act. Like it or not, it's the model that seems to be working for us.
Thank you for your interest in Lustre -- I hope you enjoy Lustre 1.2.4!