Thanks for your answer. Yes, I forgot to mention I was planning on having each node be a metadata and an I/O. OK, backup metadata too. I am not sure how I'd recover from a node problem, though; I guess I need to do a few experiments. (Two scenarios: a) node goes down, and everything gets stopped; b) node goes down, and I decide to continue working with the other).
Best, R. On Sat, 19 May 2012 22:16:11 -0400,Boyd Wilson <[email protected]> wrote: > OrangeFS works fine across the compute nodes as storage, many people use it > that way, so you are fine. As with the backup, also make sure you take > regular backups of the metadata as well. The easiest setup would be to > have each node be a metadata and an I/O node. > -boyd > On Sat, May 19, 2012 at 6:53 PM, Ramon Diaz-Uriarte <[email protected]>wrote: > > > > Dear All, > > > > I have some general questions about whether, in my setup, it makes sense > > to use OrangeFS and, if it makes sense, the recommended usage patterns. > > > > > > Context > > ======= > > > > We have a two node cluster (a Dell PowerEdge C6145). Each node has four > > SAS HDs (600 GB each) per node (which I'll most likely use as a single > > virtual disk using the RAID card). Each node also has four AMD Opteron > > 6276 sockets (16 cores per socket) and 256 GB RAM. The two nodes are > > connected via Infiniband. > > > > > > We will be using the cluster for bioinformatics/statistics computing, > > including programs that use MPI and OpenMP, as well as giving access (via > > web-based applications) to those same bioinfo/stats computing programs. > > > > > > The main reason for considering OrangeFS is to provide a single > > shared-disk file system for homes, application code, result storage, > > scratch space, tmp files, etc. > > > > > > Questions: > > ========== > > > > > > 1. I've seen it recommended that computing and storage nodes be different > > (e.g., FAQ, 3.6). However, having different computing and storage nodes > > would not make sense in my case. Anything I am missing? > > > > > > 2. Is it an overkill to use OrangeFS in my scenario? > > > > > > 3. If one of the nodes fails, and if I keep a backup of the data, can I > > recover by just copying the shared disk backup, and mounting as a > > regular, local, file system? (To allow this, I think I need to create > > the backups from one of the clients, not just the data held by a single > > server ---i.e., I would not use the approach in the FAQ, 9.1) > > > > > > 4. I understand things will be much cleaner if I use a dedicated partition > > (in each machine) to be used as a brick, instead of a directory? > > > > > > 5. What other options might I consider? I am also thinking about GlusterFS > > (and asking similar questions on their list). (Lustre definitely seems > > to discourage client and OSS in same node). > > > > > > Any other comments or suggestions for this setup are welcome. > > > > > > Best, > > > > R. > > > > -- > > Ramon Diaz-Uriarte > > Department of Biochemistry, Lab B-25 > > Facultad de Medicina > > Universidad Autónoma de Madrid > > Arzobispo Morcillo, 4 > > 28029 Madrid > > Spain > > > > Phone: +34-91-497-2412 > > > > Email: [email protected] > > [email protected] > > > > http://ligarto.org/rdiaz > > > > > > _______________________________________________ > > Pvfs2-users mailing list > > [email protected] > > http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users > > -- Ramon Diaz-Uriarte Department of Biochemistry, Lab B-25 Facultad de Medicina Universidad Autónoma de Madrid Arzobispo Morcillo, 4 28029 Madrid Spain Phone: +34-91-497-2412 Email: [email protected] [email protected] http://ligarto.org/rdiaz _______________________________________________ Pvfs2-users mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
