hi joe, it's very reassuring to read your post. answers in-line. > I do believe it is a serious and stable system. We are running into > specific bugs. I'm ok with workarounds, but I really want to find a way to > cause the problems. > same with both. right now i'm putting more of my production tree back onto an NFS server. i think this is quite a reproducible bug: just check out a subversion repo onto a distributed replicated volume - alter some files, check them back in and repeat a few times. within <20 mins we've been getting input/output errors and stale NFS handles. this is with both NFS/fuse clients on multiple clients. No one really questions other parallel FS, and from some of our experiences > with them, they have far worse issues. Even with the issues, Gluster is one > of the best on the market. > > Gluster is evolving, but the issue is, without the replicator, all we can > do is say "we are experiencing an issue" and hope it can be tracked. i hear you - and very much understand the value of reproducible bug reports :) i hope above example helps. i've already sent a detailed report to the gluster people last week. i'd be happy to give them a ssh account to see it first hand. i want to help/contribute... > i'd really like to hear from any official gluster people out there - >> right now the silence is deafening. is this issue know? is it viewed a >> > > I don't know if you have a support contract with them. If you do, you > should be speaking with them directly. If not, their paying customers come > first. i have chatted with them - and certainly would expect that to service paying customers (which we are not yet). i've had couple of chats - and they seem like a very cool helpful switched on bunch... again, thanks for responding. good to know i'm not alone.. -paul
