[轉] 關於 Ceph PG

本系列文章會深入研究 Ceph 以及 Ceph 和 OpenStack 的集成：

（1）安裝和部署

（2）Ceph RBD 接口和工具

（3）Ceph 物理和邏輯結構

（4）Ceph 的基礎數據結構

（5）Ceph 與 OpenStack 集成的實現

（6）QEMU-KVM 和 Ceph RBD 的 緩存機制總結

（7）Ceph 的基本操作和常見故障排除方法

（8）關於Ceph PGs

 

https://docs.google.com/presentation/d/1_0eIWNYWvJON4GrJwktQ1Rumgqesc8QQbVkhutXgcNs/edit#slide=id.g7365a7bef_0_17 

 

Placement Group, PG

 

How PGs are Used?

 

Should OSD#2 fail

 

Data Durability Issue

 

Object Distribution Issue

 

Choosing the number of PGs

 

Example

 

如何根據現狀調整 PG and PGP?

 

Monitoring OSDs

 

Ceph is NOT Healthy_OK

 

OSD Status Check

 

PG Sets

 

When A OSD in the Acting Set is down

 

Up Set

 

Check PG Status

 

Point

 

Peering

 

Peering: Establish Agreement of the PG status

 

Monitoring PG States

 

Check PG Stat

 

List Pool

 

PG IDs

 

The Output Format of the placement group

 

 PG

Creating PG

 

Create A Pool

 

 Peering

A Peering Process for a Pool with Replica 3

 

Active

 

Clean

 

DEGRADED

 

PG with {active + degraded}

 

Recovering

 

Backing Filling (1/2)

 

Backing Filling (2/2)

 

Remapped

 

 Stale

Identifying Troubled PGs (1/2)

 

Identify Trouble PGs (2/2)

 

Finding An Object Location