Daniel, I'm going to use glusterfs in a geodetic computing system. That gives much io, and also applications running glusterfs mounts. We have been using NFS, but it is slow and bad on big file trees, consuming much load, and bad scalability. Not recommended. Now i partly use glusterfs 1.2.3, having just a problem with too many open files. I also want network mirroring, and will maybe try Gluster +HeartBeat+DRBD with unify, just pairing the discs with DRBD, and unify on top of that. Best Einar On tor, 2007-07-05 at 11:33 -0300, Daniel van Ham Colchete wrote: > Harry, > > I'm sorry for being so nosy, but could you please clarify something about > your production environment? The project I have been planning for 8 months > will go to production in less than 3 weeks. Because of the current AFR > status I'm between Gluster+HeartBeat+DRBD or NFS+Heartbeat+DRBD (without > anything like unify, witch is a minus). In the Gluster case I'll use > mainline-2.5. > > Best regards, > Daniel Colchete > > On 7/5/07, Harris Landgarten <harrisl@xxxxxxxxxxxxx> wrote: > > > > Einar, > > > > It depends on how you are using gluster. I am running 2.5 in a production > > environment but I only use unify, posix-locks, readahead, writebehind and > > io-threads. I would not use AFR yet with 2.5 since there is still some > > debugging happening but it seems to be getting very close. Others may have > > differing opinions. > > > > Best > > > > Harris
