convoy RBD driver support

Related projects:

https://github.com/yp-engineering/rbd-docker-plugin
https://github.com/AcalephStorage/docker-volume-ceph-rbd
https://github.com/contiv/volplugin

Aug 21 '15 15:08 leseb

It seems ceph RBD would fit very well as driver. I'd love to see support for it.

Sep 13 '15 14:09 tomzo

+1 Ceph driver would be great!

Sep 26 '15 07:09 pwFoo

Any plans to add ceph support? would be important for some setups.

Oct 08 '15 12:10 pwFoo

+1

Oct 11 '15 21:10 tlvenn

Another ceph plugin, but with last commit 7 months ago: rbd-volume - NON-FUNCTIONAL https://github.com/ceph/ceph-docker/tree/master/rbd-volume

Oct 15 '15 19:10 pwFoo

+1 For those wanting to try ceph for docker without convoy, look at the page below, using https://github.com/yp-engineering/rbd-docker-plugin . I tested on docker 1.9.0 where it seems to work fine.

http://www.sebastien-han.fr/blog/2015/08/17/getting-started-with-the-docker-rbd-volume-plugin/

Nov 04 '15 16:11 ps-account

@pimpim please let us know how you like it and if you need any features from it.

Nov 06 '15 00:11 hekaldama

@hekaldama I am currently trying out the yp-engineering with a proof-of-concept ceph installation, no authentication or users inside of the ceph cluster. Maybe I should give some feedback on their project page as well, but since you ask for it:

I really like the plugin's ability for flexible creating of volumes of a certain size in ceph: e.g. the "@512" option to create a 512 MB sized volume works great, could also make volumes larger than the default setting (in this case 20GB), maybe nicer to even use a M/G/T suffix to allow creating volumes between mega and terabytes without getting crazy long numbers.

I also like that I can load several plugins per host, so I can use different pools simultaneously within the same docker instance. I wonder if you could even get pools from different ceph clusters, but then the host should be a client of different ceph clusters, not sure if that is possible.

The options to delete a scratch volume directly after use or rename them for later cleanup look promising.

Something I would like to see, but maybe that is just my weird use case - if there are different users on a docker host and each user corresponds to a different ceph user, it would be nice to allow the authentication at the user level during " docker run". Currently, all users that have access to the docker host basically share the access credentials to the ceph pools as given during the loading of the plugin on the host. But that's a similar issue that NFS and CIFS plugins have.

Nov 09 '15 12:11 ps-account

@pimpim added some tickets(https://github.com/yp-engineering/rbd-docker-plugin/issues/11, https://github.com/yp-engineering/rbd-docker-plugin/issues/12) to our project for your suggested enhancements.

As far auth during docker run, sounds like you are saying similar thing as https://github.com/yp-engineering/rbd-docker-plugin/issues/10? If so, please comment on that ticket with any other suggestions and such.

Thanks.

Nov 09 '15 12:11 hekaldama

+1, its a must for us, we currently have to pin services to hosts.