On 10/02/2013 03:16 PM, Blair Bethwaite wrote:
Hi Josh,Message: 3 Date: Wed, 02 Oct 2013 10:55:04 -0700 From: Josh Durgin <josh.durgin@xxxxxxxxxxx> To: Oliver Daudey <oliver@xxxxxxxxx>, ceph-users@xxxxxxxxxxxxxx, Robert.vanLeeuwen@xxxxxxxxxxxxx Subject: Re: Loss of connectivity when using client caching with libvirt Message-ID: <524C5DF8.6000503@xxxxxxxxxxx> Content-Type: text/plain; charset=ISO-8859-1; format=flowed The behavior you both are seeing is fixed by making flush requests asynchronous in the qemu driver. This was fixed upstream in qemu 1.4.2 and 1.5.0. If you've installed from ceph-extras, make sure you're using the .async rpms [1] (we should probably remove the non-async ones at this point). The cuttlefish qemu rpms should work fine with dumpling. They're only separate from the bobtail ones to be able to use newer functions in librbd.The OP piqued my interest with this as we are looking at caching options on Ubuntu Precise (Ceph and Cloud) with Dumpling. Do the same caveats apply for qemu-kvm on Precise? Presumably with just read caching there is no such problem?
The version base of qemu in precise has the same problem. It only affects writeback caching. You can get qemu 1.5 (which fixes the issue) for precise from ubuntu's cloud archive. Josh _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
