bug: ceph pool create --size 1fail

[benoitc]

I get the following error when trying to create a pool with only 1 copy:

pveceph pool create readers --min_size 1 --size 1 --pg_num 512 mon_command failed - pg_num 512 size 3 would mean 1920 total pgs, which exceeds max 1500 (mon_max_pg_per_osd 250 * num_in_osds 6)

It seems to default to 3.

 

[benoitc]

the idea behind was to apply part of the Deploying MySQL databases on Redhat Ceph Storage whitepaper

When MySQL replication is used in conjunction with Ceph RBD volumes instead, only the leader would have a full copy of the dataset, and the followers would operate on a thin-provisioned clone of the leader’s dataset. Furthermore, while the leader’s dataset would be fully replicated to all storage nodes, the follower clones would have only one copy. A leader failover simply moves the leader server and its associated Ceph block device to another node. Since the follower clones are copy-on-write and ephemeral, disk capacity can be saved by reprovisioning followers at a semi-regular interval.​

SO I thought to have ephemeral vms for it. But I am not sure how proxmox manage ephemeral vms with only 1 copy though. Thoughts?

 

[Alwin]

mon_command failed - pg_num 512 size 3 would mean 1920 total pgs, which exceeds max 1500 (mon_max_pg_per_osd 250 * num_in_osds 6)

Set the sizes and pgs after you created the pool with ceph's tooling. In the current version the pveceph doesn't have a set command yet.