Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Massimiliano Cuttini
Not really tested, but searching around, many people say at moment RBD-NBD as more or less same perforamance. While RBD-FUSE is really slow. At the moment I cannot anymore test the kernel version because downgrading/re-upgrading CRUSH tunable will be a nightmare. But you can try. Il 27/

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Massimiliano Cuttini
But if everybody get Kernel Mismatch (me too) ... why don't use directly rbd-nbd and forget about kernel-rbd All feature, almost same performance. No? Il 27/02/2017 18:54, Ilya Dryomov ha scritto: On Mon, Feb 27, 2017 at 6:47 PM, Shinobu Kinjo wrote: We already discussed this: https

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Ilya Dryomov
On Mon, Feb 27, 2017 at 6:47 PM, Shinobu Kinjo wrote: > We already discussed this: > > https://www.spinics.net/lists/ceph-devel/msg34559.html > > What do you think of comment posted in that ML? > Would that make sense to you as well? Sorry, I dropped the ball on this. I'll try to polish and push

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Vasu Kulkarni
thanks for that link, It will be nice to have that interface supported by .ko, regardless i raised this http://tracker.ceph.com/issues/19095 On Mon, Feb 27, 2017 at 9:47 AM, Shinobu Kinjo wrote: > We already discussed this: > > https://www.spinics.net/lists/ceph-devel/msg34559.html > > What do

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Shinobu Kinjo
We already discussed this: https://www.spinics.net/lists/ceph-devel/msg34559.html What do you think of comment posted in that ML? Would that make sense to you as well? On Tue, Feb 28, 2017 at 2:41 AM, Vasu Kulkarni wrote: > Ilya, > > Many folks hit this and its quite difficult since the error

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Vasu Kulkarni
Ilya, Many folks hit this and its quite difficult since the error is not properly printed out(unless one scans syslogs), Is it possible to default the feature to the one that kernel supports or its not possible to handle that case? Thanks On Mon, Feb 27, 2017 at 5:59 AM, Ilya Dryomov wrote: >

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Ilya Dryomov
On Mon, Feb 27, 2017 at 3:15 PM, Simon Weald wrote: > Hi Ilya > > On 27/02/17 13:59, Ilya Dryomov wrote: >> On Mon, Feb 27, 2017 at 2:37 PM, Simon Weald wrote: >>> I've currently having some issues making some Jessie-based Xen hosts >>> talk to a Trusty-based cluster due to feature mismatch error

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Simon Weald
Hi Ilya On 27/02/17 13:59, Ilya Dryomov wrote: > On Mon, Feb 27, 2017 at 2:37 PM, Simon Weald wrote: >> I've currently having some issues making some Jessie-based Xen hosts >> talk to a Trusty-based cluster due to feature mismatch errors. Our >> Trusty hosts are using 3.19.0-80 (the Vivid LTS ker

Re: [ceph-users] krbd and kernel feature mismatches

2017-02-27 Thread Ilya Dryomov
On Mon, Feb 27, 2017 at 2:37 PM, Simon Weald wrote: > I've currently having some issues making some Jessie-based Xen hosts > talk to a Trusty-based cluster due to feature mismatch errors. Our > Trusty hosts are using 3.19.0-80 (the Vivid LTS kernel), and our Jessie > hosts were using the standard