Dear all, This maybe one of the FAQs on OCFS. From Oracle's note : 237997.1 "Linux OCFS - Best Practice" point 5 "Calculating an OCFS Partition Size" it mentions: OCFS supports partitions of up to 1Tb (tested). Since no volume management is built into OCFS, Oracle recommends enabling hardware raid support to create logical disk volumes of sufficient size. If hardware raid support is not available, a Logical Volume Manager (LVM) or Multi-Disk (MD) disk configuration can be employed depending on Linux distribution being used. ------------- end quote------------- And also from Oracle's note : 224586.1 "Oracle Cluster File System (OCFS) on RedHat AS - FAQ" : *Q: 29. Can I create an OCFS filesystem over a LVM (Logical Volume Manager)?* *A:* No, LVM is not cluster aware, No tests have been performed using LVM, It may work, but it also may not. Corruptions are likely to happen since they are not cluster aware software. ------------- end quote------------- Seems a bit contradiction, and it also seems quite usual and normal for large database to have logical volumes created from several harddisks. Does it work ? If so, is it supported? Thanks. cheers, geoff
> LVM is not cluster aware, No tests have been performed using LVM, It may > work, but it also may not. > Corruptions are likely to happen since they are not cluster aware software. > > ------------- end quote------------- > > Seems a bit contradiction, and it also seems quite usual and normal for > large database to have logical volumes created from several harddisks. > Does it work ? If so, is it supported?not sure what contradicts. 1- hardware raid/volume mgmt works, most folks use that, no problem 2- "lvm is not clusteraware" . if you want to use lvm, there is nothing Today that works perfectly clean on linux, in a cluster, so, does it work ? mostly. is it supported ? well, if you rvolume manager screws up its not an ocfs problem. ocfs by itself does not need anything special, if you were to find a volume manager that can be trusted to behave correctly in a clutser, go ahead, but if you resize a volume on one node, and the other nodes don't know about this and overwrite stuff, well, you 're on your own.