Not having run it for a long time I don't feel we can make a great long-term judgement. We are looking at this solution for reasons of reliability. The feeling is while our lustre filesystems provide the best performance and used for scratch space on a cluster they are a little unstable (or untrusted) to be used for longer term data storage. Using gluster without striping leaves a very simple backend system with normal files - aiding in the case filesystem recovery is needed. During our testing and working through various problems we have greatly abused the gluster install in general, and the data has remained fine throughout. So that part of it looks good. The people here working on the kernel specific issues are still in the process, when they're done any findings will get posted to the list. Long term, I'm still concerned about the lack of documentation. Scott On 8/4/11 8:51 AM, Uwe Kastens wrote: > Scott, > > That sound interesting. > > It hasn't been going well so far. Currently we are struggling with > performance problems that seem to be linked to certain versions of > the linux kernel. > > The lack of in depth and current documentation is one of the most > glaring and obvious problems. > > > > What is your feeling about stability of the solution? > > Kind Regards > > Uwe > > > > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
