proxmox超融合私有云发布最新版本PVE 5.3

奥地利维也纳，2018年12月4日，proxmox官方正式发布PVE5.3版本。

Proxmox Server Solutions GmbH today unveiled Proxmox VE 5.3, its latest open-source server virtualization management platform. Proxmox VE is based on Debian Stretch 9.6 with a modified Linux Kernel 4.15. Ceph Storage has been updated to version 12.2.8 (Luminous LTS, stable), and is packaged by Proxmox.

proxmox公司发布最新的开源虚拟化管理平台PVE5.2.PVE基于流行的linux发行版Debian Stretch 9.6，内核版本4.15.去中心化存储ceph已经更新到Luminous 12.2.8版本。该版本由proxmox重新封装。

Proxmox VE and CephFS

Proxmox VE 5.3 now includes CephFS in its web-based management interface thus expanding its comprehensive list of already supported file and block storage types. CephFS is a distributed, POSIX-compliant file system and builds on the Ceph cluster. Like Ceph RBD (Rados Block Device), which is already integrated into Proxmox VE, CephFS now serves as an alternative interface to the Ceph storage. For CephFS Proxmox allows storing VZDump backup files, ISO images, and container templates. The distributed file system CephFS eliminates the need for external file storage such as NFS or Samba and thus helps reducing hardware cost and simplifies management.

PVE5.3 默认包含Cephfs,可以直接用web界面进行配置和管理（注：pve5.2版本用 pveceph install安装，用命令行初始化pveceph init --network网络).cephfs支持文件及块存储。cephfs是分布式的、兼容POSIX、去中心化的集群。cephfs作为存储服务接口，可以存储虚拟机备份文件、ISO镜像（注：操作系统镜像共享可以在任意节点进行读取，不过把这些不使用频率低的ISO放在ceph存储，似乎有点浪费资源，译者本人一般使用独立的低速NFS来存储）、容器模板等。cephFS可以完全取代传统的外挂存储如NFS、Samba等等，降低固定资产投入及简化架构和管理。

The CephFS file system can be created and configured with just a few clicks in the Proxmox VE management interface. To deploy CephFS users need a working Ceph storage cluster and a Ceph Metadata Server (MDS) node, which can also be created in the Proxmox VE interface. The MDS daemon separates metadata and data from each other and stores them in the Ceph file system. At least one MDS is needed, but its recommended to deploy multiple MDS nodes to improve high availability and avoid SPOF. If several MDS nodes are created only one will be marked as ‘active’ while the others stay ‘passive’ until they are needed in case of failure of the active one.
在proxmox VE web管理界面，点几下鼠标就能完成文件系统的创建和配置（译者注：以前的版本是在命令行完成，直到创建监视器）。以web管理界面创建ceph的元数据及实际数据存储（译者注：创建OSD后，用fdisk命令可以看到物理磁盘被分割成两个分区，大小为100M的分区为元数据MDS），守护进程MDS把元数据跟实际数据分离（不吃大锅饭，提高效率）。在一个ceph集群中，至少需要一个元数据服务器，为了保证可靠性，建议使用多个MDS。虽然有多个元数据服务器，但正常情况下，只有一个元数据服务器处于活跃状态，如果管事的元数据服务器故障，其它服务器选举新的一个来接替。

补充：升级办法

1、PVE web界面升级

自动弹出窗口，输入字母“y"继续。

2、命令行方式：
登录没一个节点，执行apt-get update && apt-get upgrade -y

升级完毕，所需服务自动重启，无需手动重启操作系统。