Design questions
shulemmosko
33 Posts
February 7, 2021, 11:03 pmQuote from shulemmosko on February 7, 2021, 11:03 pmHi all.
We currently run a Petasan cluster of 3 nodes, each node has 6x 1.92Tb enterprise SAS SSD drives the pool is replicated 3 times and used for running VM's.
Question: We want to have Ceph s3 capabilities but not use these expensive drives for it, can we add a fourth server with regular HDD and create a separate ec pool of these drives and s3 to use only the ec pool?
the s3 pool is mainly going to be for backup storage.
How should we go about this setup?
Thank!!
Shulem
Hi all.
We currently run a Petasan cluster of 3 nodes, each node has 6x 1.92Tb enterprise SAS SSD drives the pool is replicated 3 times and used for running VM's.
Question: We want to have Ceph s3 capabilities but not use these expensive drives for it, can we add a fourth server with regular HDD and create a separate ec pool of these drives and s3 to use only the ec pool?
the s3 pool is mainly going to be for backup storage.
How should we go about this setup?
Thank!!
Shulem
admin
2,930 Posts
February 8, 2021, 8:47 amQuote from admin on February 8, 2021, 8:47 amyes you can create a crush rule to place data on this 4th node only, then create ec pool with this rule. note that if this node itself fails, your pool will not be active. it is more common to create a slower hdd pool on a cluster of nodes, either separate new nodes or on the same nodes serving ssd pool with different crush rules with different device class, we have example crush templates for this in the ui.
yes you can create a crush rule to place data on this 4th node only, then create ec pool with this rule. note that if this node itself fails, your pool will not be active. it is more common to create a slower hdd pool on a cluster of nodes, either separate new nodes or on the same nodes serving ssd pool with different crush rules with different device class, we have example crush templates for this in the ui.
Design questions
shulemmosko
33 Posts
Quote from shulemmosko on February 7, 2021, 11:03 pmHi all.
We currently run a Petasan cluster of 3 nodes, each node has 6x 1.92Tb enterprise SAS SSD drives the pool is replicated 3 times and used for running VM's.
Question: We want to have Ceph s3 capabilities but not use these expensive drives for it, can we add a fourth server with regular HDD and create a separate ec pool of these drives and s3 to use only the ec pool?
the s3 pool is mainly going to be for backup storage.
How should we go about this setup?
Thank!!
Shulem
Hi all.
We currently run a Petasan cluster of 3 nodes, each node has 6x 1.92Tb enterprise SAS SSD drives the pool is replicated 3 times and used for running VM's.
Question: We want to have Ceph s3 capabilities but not use these expensive drives for it, can we add a fourth server with regular HDD and create a separate ec pool of these drives and s3 to use only the ec pool?
the s3 pool is mainly going to be for backup storage.
How should we go about this setup?
Thank!!
Shulem
admin
2,930 Posts
Quote from admin on February 8, 2021, 8:47 amyes you can create a crush rule to place data on this 4th node only, then create ec pool with this rule. note that if this node itself fails, your pool will not be active. it is more common to create a slower hdd pool on a cluster of nodes, either separate new nodes or on the same nodes serving ssd pool with different crush rules with different device class, we have example crush templates for this in the ui.
yes you can create a crush rule to place data on this 4th node only, then create ec pool with this rule. note that if this node itself fails, your pool will not be active. it is more common to create a slower hdd pool on a cluster of nodes, either separate new nodes or on the same nodes serving ssd pool with different crush rules with different device class, we have example crush templates for this in the ui.