Comments for Ceph http://ceph.com Fri, 24 May 2013 11:23:54 +0000 hourly 1 http://wordpress.org/?v=3.3.2 Comment on What’s New in the Land of OSD? by Igor Laskovy http://ceph.com/dev-notes/whats-new-in-the-land-of-osd/#comment-22698 Igor Laskovy Fri, 24 May 2013 11:23:54 +0000 http://ceph.com/?p=2083#comment-22698 Awesome explanation! Reqesting a new part of "waht's new in" for cuttlefish reaease please! ;) Awesome explanation!

Reqesting a new part of “waht’s new in” for cuttlefish reaease please! ;)

]]> Comment on Adding Support for RBD to stgt by dmick http://ceph.com/dev-notes/adding-support-for-rbd-to-stgt/#comment-20988 dmick Tue, 14 May 2013 19:48:08 +0000 http://ceph.com/?p=3177#comment-20988 Hi Christopher! Thanks for reading. At the moment, Ceph doesn't take advantage of RDMA per se, although we've seen good performance on IPoIB. It's possible rsockets will gain us some of the advantages of RDMA without much code change, but we haven't had time to set up that experiment just yet (although we hope to). But there's no reason you couldn't run tgtd to connect to ceph over IPoIB today. Please feel free to join us on irc.oftc.net #ceph, or <a href="http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com" rel="nofollow">the ceph-users mailing list</a>, for more discussion. Hi Christopher! Thanks for reading. At the moment, Ceph doesn’t take advantage of RDMA per se,
although we’ve seen good performance on IPoIB. It’s possible rsockets will gain us some of the
advantages of RDMA without much code change, but we haven’t had time to set up that experiment
just yet (although we hope to). But there’s no reason you couldn’t run tgtd to connect to ceph
over IPoIB today.

Please feel free to join us on irc.oftc.net #ceph, or the ceph-users mailing list, for more discussion.

]]> Comment on Adding Support for RBD to stgt by Christopher Barry http://ceph.com/dev-notes/adding-support-for-rbd-to-stgt/#comment-20684 Christopher Barry Mon, 13 May 2013 00:45:09 +0000 http://ceph.com/?p=3177#comment-20684 I'm using tgtd with Infiniband and iser now, connecting to arrays of SSDs from Linux hypervisors, which then export these devices to kvm virtual machines. How would I best exploit tgtd and RADOS over Infiniband using RDMA? I’m using tgtd with Infiniband and iser now, connecting to arrays of SSDs from Linux hypervisors, which then export these devices to kvm virtual machines. How would I best exploit tgtd and RADOS over Infiniband using RDMA?

]]> Comment on Ceph Over Fibre for VMWare by Jared http://ceph.com/community/ceph-over-fibre-for-vmware/#comment-20264 Jared Thu, 09 May 2013 22:34:23 +0000 http://ceph.com/?p=2107#comment-20264 I am investigating using Ceph as a storage target for virtual servers in VMware. We have 3 servers packed with hard drives ready for the proof of concept. I am looking for some direction. Do you have any documentation you can share? Any help, would be greatly appreciated. I am investigating using Ceph as a storage target for virtual servers in VMware. We have 3 servers packed with hard drives ready for the proof of concept. I am looking for some direction. Do you have any documentation you can share? Any help, would be greatly appreciated.

]]> Comment on Ceph at World Hosting Days 2013 by Lennie http://ceph.com/events/ceph-at-world-hosting-days-2013/#comment-15861 Lennie Wed, 03 Apr 2013 13:15:09 +0000 http://ceph.com/?p=3215#comment-15861 I do have a small tip: the person hosting the panel talked about 'cloud storage'. That isn't a very precise term, I'm afraid a lot of people have their own idea of what that means. I do have a small tip: the person hosting the panel talked about ‘cloud storage’. That isn’t a very precise term, I’m afraid a lot of people have their own idea of what that means.

]]> Comment on Ceph at World Hosting Days 2013 by scuttlemonkey http://ceph.com/events/ceph-at-world-hosting-days-2013/#comment-15858 scuttlemonkey Wed, 03 Apr 2013 12:24:34 +0000 http://ceph.com/?p=3215#comment-15858 Yep, sure did. Should be fixed now, thanks! Yep, sure did. Should be fixed now, thanks!

]]> Comment on v0.60 released by scuttlemonkey http://ceph.com/releases/v0-60-released/#comment-15857 scuttlemonkey Wed, 03 Apr 2013 12:24:06 +0000 http://ceph.com/?p=3243#comment-15857 Yep, missed the 0. in the url. Should be fixed now, thanks! Yep, missed the 0. in the url. Should be fixed now, thanks!

]]> Comment on Ceph at World Hosting Days 2013 by Lennie http://ceph.com/events/ceph-at-world-hosting-days-2013/#comment-15856 Lennie Wed, 03 Apr 2013 12:20:43 +0000 http://ceph.com/?p=3215#comment-15856 Maybe I'm wrong, but I think you meant "high performance computing" instead of "high power computing". Maybe I’m wrong, but I think you meant “high performance computing” instead of “high power computing”.

]]> Comment on v0.60 released by Bram Pieters http://ceph.com/releases/v0-60-released/#comment-15850 Bram Pieters Wed, 03 Apr 2013 09:31:35 +0000 http://ceph.com/?p=3243#comment-15850 The tarball link seems to be broken. The tarball link seems to be broken.

]]> Comment on CephFS MDS Status Discussion by Vladimir http://ceph.com/dev-notes/cephfs-mds-status-discussion/#comment-14439 Vladimir Mon, 18 Mar 2013 07:27:24 +0000 http://ceph.com/?p=3042#comment-14439 In response to Adam: ceph -w is unable to show remaining disk space taking into account replication level for a simple reason: ceph may have number of pools with different replication levels. Say you have four racks filled with OSDs and you would like to make sure that a copy of precious data on each rack. So you make a pool named 'precious' with replication level 4. But in order to run interim data faster you create another pool named 'fastdata' with replication level 2. What ceph -w should show then? It may be good to have % capacity used but it will clutter ceph -w output. I personally use cacti to grab used and total figures, put it on graph and set out nagios warning if used capacity exceeds 80% - that's a time when I start to look for new OSD or extra drives. Why space has gone: issue of storage users and not of storage itself. We pretty much declared to our users - space is unlimited and just will throw in extra hard drives as needed. In response to Adam:

ceph -w is unable to show remaining disk space taking into account replication level for a simple reason: ceph may have number of pools with different replication levels. Say you have four racks filled with OSDs and you would like to make sure that a copy of precious data on each rack. So you make a pool named ‘precious’ with replication level 4. But in order to run interim data faster you create another pool named ‘fastdata’ with replication level 2. What ceph -w should show then?

It may be good to have % capacity used but it will clutter ceph -w output. I personally use cacti to grab used and total figures, put it on graph and set out nagios warning if used capacity exceeds 80% – that’s a time when I start to look for new OSD or extra drives. Why space has gone: issue of storage users and not of storage itself. We pretty much declared to our users – space is unlimited and just will throw in extra hard drives as needed.

]]>