A note about non-enterprise flash storage:
Running Ceph without Power Loss Protection (PLP) on SSDs, especially consumer drives, is risky: it can cause data corruption because writes aren't fully committed before power loss, and slow performance and high wear; enterprise drives with PLP use capacitors to safely flush data, preventing these issues, making PLP crucial for Ceph's synchronous writes. While some home labs might tolerate it, it's a recipe for disaster in production, so using PLP-enabled enterprise SSDs is highly recommended.
Risks of no Power Loss Protection (PLP)
- Data Corruption: Unflushed data in SSD caches can be lost during power cuts.
- Performance Degradation: Without PLP, the SSD must write directly to NAND for every sync write (like Ceph's), slowing down operations significantly compared to enterprise drives that use cache and flush later.
- High Write Amplification (WA): Ceph's synchronous writes hit consumer drives hard, increasing internal SSD wear (TBW) and shortening their lifespan.
Show all PGs
ceph pg dump
Show OSDs are within the Up Set and Acting Set
ceph pg map {pg-num}