Hi, > On 31 Jan 2022, at 00:53, Nir Soffer <nsoffer@xxxxxxxxxx> wrote: > > Live migration and snapshots are not available? This is news to me. > Welcome to krbd world. But there will be no fun when you need to update the rbd driver for the entire aggregate with migration and reboot for each host > Why do you need to connect 300-400 rbd devices at the same time to a host? Because projects are already run and this is projects scale/maintenance path > Few years ago we decided to replace the incomplete, unmaintained, and untested > code with a better way - using cinderlib - called Managed Block Storage (MBS). But at the same time, support was dropped in a minor release of 4.4.6, and not in 4.5 in which it was planned. This is fine? I think it's doubtful > Do you have 400 active disks actually used by VMs on every host? 400 no, but 300 - yes, this is only 8 VMS with 41 disk's, 4CPU 8GB RAM instances - the Gold CPU's can handle even more... https://ibb.co/HHJzkZX <https://ibb.co/HHJzkZX> screenshot > If you do, do they perform worse compared with 400 LUNs, or 400 active > logical volumes? I don't know what is it, we are use "old Cinder oVirt integration (from 2016)" and it is work flawless <disk type='network' device='disk' snapshot='no'> <driver name='qemu' type='raw' cache='writeback' error_policy='stop' io='threads' discard='unmap'/> <auth username='cinder'> <secret type='ceph' uuid='e0828f39-2832-4d82-90ee-23b26fc7b20a'/> </auth> <source protocol='rbd' name='replicated_rbd/volume-86769788-7824-4809-bc56-20ef5cee18fa' index='37'> <host name='172.16.16.2' port='3300'/> <host name='172.16.16.3' port='3300'/> <host name='172.16.16.4' port='3300'/> </source> <target dev='sda' bus='scsi'/> <serial>86769788-7824-4809-bc56-20ef5cee18fa</serial> <boot order='1'/> <alias name='ua-86769788-7824-4809-bc56-20ef5cee18fa'/> <address type='drive' controller='0' bus='0' target='0' unit='0'/> </disk> <disk type='network' device='disk' snapshot='no'> <driver name='qemu' type='raw' cache='writeback' error_policy='stop' io='threads' discard='unmap'/> <auth username='cinder'> <secret type='ceph' uuid='e0828f39-2832-4d82-90ee-23b26fc7b20a'/> </auth> <source protocol='rbd' name='replicated_rbd/volume-9edd06d5-b0cc-437f-9aaa-80ccc391f302' index='36'> <host name='172.16.16.2' port='3300'/> <host name='172.16.16.3' port='3300'/> <host name='172.16.16.4' port='3300'/> </source> <target dev='sdb' bus='scsi'/> <serial>9edd06d5-b0cc-437f-9aaa-80ccc391f302</serial> <alias name='ua-9edd06d5-b0cc-437f-9aaa-80ccc391f302'/> <address type='drive' controller='0' bus='0' target='0' unit='1'/> </disk> Thanks, k _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |