Hi Mike,
Have you considered creating SR which doesn't make one huge RBD volume
and on top of it creates LVM but instead creates separate RBD volumes
for each VDI?
W dniu 25.02.2017 o 22:14, Mike Jacobacci pisze:
Hi Max,
I have a working Xenserver pool using Ceph RBD as a backend.. I got it
working by using the RBDSR plugin here:
https://github.com/mstarikov/rbdsr
I don't have much time, but I just wanted to respond in case it's
helpful... Here is how I got it working:
On CEPH
Set tunables to legacy
# ceph osd crush tunables legacy
Disable chooseleaf_vary_r
# ceph osd getcrushmap -o /tmp/mycrushmap
# crushtool -d /tmp/mycrushmap > /tmp/mycrushmap.txt
chooseleaf_vary_r 0
# crushtool -c /tmp/mycrushmap.txt -o /tmp/mycrushmap.new
# ceph osd setcrushmap -i /tmp/mycrushmap.new
Create RBD image
# rbd create */pool/*//*image*/ --size *512G* --image-shared
Remove Deep-flatten feature
# rbd feature disable */pool/*/*/image/* deep-flatten
On XenServer
(Install the RBDSR plugin on every node)
To Create SR on XenServer
# modprobe rbd (make sure the are no errors)
# xe sr-create type=lvmoiscsi name-label=/*RBD IMAGE NAME*/
shared=true device-config:target=*/CEPH MONITOR IP/*
device-config:port=6789 device-config:targetIQN=*/RBD POOL NAME/*
device-config:SCSIid=*/MOUNT NAME/*
device-config:chapuser=*/CEPHUSER/*
device-config:chappassword=*/PASSWORD/*
NOTES:
- The GUI option doesn't work, but the above command works every time.
- If you you have more than one XenServer and they are on a Pool, you
have to run the sr-create command on one node, then the other(s) pool
members will need to rejoin the Pool for the sr to be shared accross
the pool. I couldn't get the mounts working on other pool members of
an existing pool.
Let me know if you have any questions.
Cheers,
Mike
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
--
Regards,
Michał Chybowski
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
