On Sun, Jul 3, 2011 at 10:33 AM, Vineet Jain <vinjvinj@xxxxxxxxx> wrote: > Also, I have several servers in my cluster. Is it ok to have the ceph > kernel client and ceph store nodes running on the same machine. Since > I have 4 cores on each box? The problem isn't the number of cores. It's the feedback loop. The classic feedback loop for NFS goes something like this: 1. Unused memory is low, so the system tries to flush the page cache for the NFS-mounted filesystem 2. The NFS server needs memory to handle the traffic from localhost. 3. go to step 1 For Ceph, you can substitute "MDS server" for NFS server. We have discussed this a lot on the list. The easiest solution is just to put the kernel client on a different box than the servers. Second easiest is to use the cfuse client, which doesn't use page cache in the same way. regards, Colin -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
