Ceph Cluster different OSD sizes - performance/issues expected?

[encore]

Hi,

Can I expect degraded performance or other issues if my CEPH cluster contains different sizes, e.g. 50x 3840GB Kingston DC1500M, and 50x7680GB Kingston DC1500M SSDs distributed over 5 nodes?

 

[itNGO]

Hi,
no if you configure CrushMap and Weight correctly there is not much of an impact expected.
CEPH Docs are your friend....

 

[encore]

aren't this values are more or less set by proxmox?
In my example at least, weight is set to the size of the disk and reweight is set to 1 for all osds, see image.

 

[itNGO]

Hi,
you can change this using the CEPH CLI and optimize your placement, so it better fits to different disk sizes.
CEPH Documentation will help you. Proxmox itself does not do much with Ceph except creating a default config with default values.