Hello, I'm in the process of doubling the number of OSD nodes in my Nautilus cluster - from 3 to 6. Based on answers receive from earlier posts to this list, the new nodes have more NVMe that the old nodes. More to the point, on the original nodes the amount of NVMe allocated to each OSD was about 120GB, so the RocksDB was limited to 30GB. However, for my workload 300GB is probably recommended. As I prepare to lay out the NVMe on these new nodes, I'm still trying to understand how to size the DB and WAL for my OSDs and whether Journal is even needed. According to https://docs.ceph.com/en/nautilus/ceph-volume/lvm/prepare/ <https://docs.ceph.com/en/nautilus/ceph-volume/lvm/prepare/>
Bluestore supports the following configurations:
* A block device, a block.wal, and a block.db device * A block device and a block.wal device * A block device and a block.db device * A single block device
First question: On my first nodes I managed to get a DB, but no WAL. My current perception is that WAL and DB occupy separate physical/logical partitions. By specifying a WAL size and a DB size, ceph-volume will create the corresponding logical volumes on the NVMe. Is this correct? It is also possible to lay these out as basic logical partitions? Second question: How do I decide whether I need WAL, DB, or both? Third question: Once I answer the above WAL/DB question, what are the guidelines for sizing them? Thanks. -Dave -- Dave Hall Binghamton University