I think one reason that they have so many stability issues is that they
are targeting at a too broad a range of hardware/software
configurations. I think at least they should have an officially
supported hardware/software configuration that is thoroughly tested and
benchmarked and have two or three officially supported volume files
targeted at a few major use cases and have the parameters properly
tuned. Life will be much easier for both the developer and the user if
less choices are left to the user. I always believe that a filesystem
is only as good as its benchmark and one that is not thoroughly
benchmarked is worse than nothing to the end user.
I'm still very interested in glusterfs as an open source project, but I
assume they need to make money out of it, and given the current economic
environment, I think they'll quickly go bankrupt if they don't do
something professional.
- Wei
Greg wrote:> One year I'm testing GlusterFS from version 1.1.4, four times I put it
> on production environment and four times it crashed (latest 2.0.6),
> it's enough, I'm leaving.
>
> Unfortunately, thanks to this project, I have now the funds to buy a
> 30K? SAN ....
>
> This message to say one more time that you have to work on stability
> before any other things (features, performance, ...)
>
> Ciao.
>