I have been reading through the documentation but have some additional questions related to our setup. We are looking to replace our Xsan (ADIC StorNext) installation with PVFS2 and want to make sure it is a good fit.
Right now we have 40 Xsan I/O servers and 2 redundant Xsan Metadata controllers connected to 20 Tbytes of Xraid storage with one large file system created over 12 LUN's using the Xsan software. Each Xsan I/O node is exporting the Xsan file system over NFS to 38 Xserve boxes running OSX 10.4. We have a total of 1520 NFS clients mounting to the 40 Xsan I/O nodes for our production system. My test configuration has 4 OpenSuse 10.1 servers SAN connected to one Xraid box configured with 2 LUNs. There are a total of 128 NFS OSX 10 clients that will access the PVFS file system over NFS. I want to reserve one of the Suse servers as a PVFS2 metadata controller and the other three as PVFS2 I/O servers exporting the PVFS2 file system to the OSX clients. As you can see I want to create a similar configuration to our existing Xsan installation. I have the PVFS server software and Kernel modules compiled and loaded on the four servers and each one sees the 2 Xraid LUNs as /dev/sdb and /dev/sdc. Can I setup the PVFS I/O servers to perform any I/O operations directly to the devices over the SAN and not the GigE network? Do I have to create a PVFS file system on the SAN device or can I use native Linux partitioning? I will continue reading and learning more about PVFS2 but some of you more experienced users can provide some insight in the meantime. Thanks in advance.
_______________________________________________ Pvfs2-users mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
