如何从 Ceph (Luminous) 集群中安全移除 OSD

简介: OSD.png工作中需要从 Ceph 的集群中移除一台存储服务器,挪作他用。Ceph 存储空间即使在移除该存储服务器后依旧够用,所以操作是可行的,但集群已经运行了很长时间,每个服务器上都存储了很多数据,在数据无损的情况下移除,看起来也不简单。
img_5fcdc033c658a988ecb31b62cf9dd241.png
OSD.png

工作中需要从 Ceph 的集群中移除一台存储服务器,挪作他用。Ceph 存储空间即使在移除该存储服务器后依旧够用,所以操作是可行的,但集群已经运行了很长时间,每个服务器上都存储了很多数据,在数据无损的情况下移除,看起来也不简单。

1. OSD 布局

先来看看 OSD 的布局

$ ceph osd tree
ID CLASS WEIGHT    TYPE NAME      STATUS REWEIGHT PRI-AFF 
-1       265.25757 root default                           
-5       132.62878     host osd7                          
24   hdd   5.52620         osd.24     up  1.00000 1.00000 
25   hdd   5.52620         osd.25     up  1.00000 1.00000 
26   hdd   5.52620         osd.26     up  1.00000 1.00000 
27   hdd   5.52620         osd.27     up  1.00000 1.00000 
28   hdd   5.52620         osd.28     up  1.00000 1.00000 
29   hdd   5.52620         osd.29     up  1.00000 1.00000 
30   hdd   5.52620         osd.30     up  1.00000 1.00000 
31   hdd   5.52620         osd.31     up  1.00000 1.00000 
32   hdd   5.52620         osd.32     up  1.00000 1.00000 
33   hdd   5.52620         osd.33     up  1.00000 1.00000 
34   hdd   5.52620         osd.34     up  1.00000 1.00000 
35   hdd   5.52620         osd.35     up  1.00000 1.00000 
36   hdd   5.52620         osd.36     up  1.00000 1.00000 
37   hdd   5.52620         osd.37     up  1.00000 1.00000 
38   hdd   5.52620         osd.38     up  1.00000 1.00000 
39   hdd   5.52620         osd.39     up  1.00000 1.00000 
40   hdd   5.52620         osd.40     up  1.00000 1.00000 
41   hdd   5.52620         osd.41     up  1.00000 1.00000 
42   hdd   5.52620         osd.42     up  1.00000 1.00000 
43   hdd   5.52620         osd.43     up  1.00000 1.00000 
44   hdd   5.52620         osd.44     up  1.00000 1.00000 
45   hdd   5.52620         osd.45     up  1.00000 1.00000 
46   hdd   5.52620         osd.46     up  1.00000 1.00000 
47   hdd   5.52620         osd.47     up  1.00000 1.00000 
-3       132.62878     host osd8                          
 0   hdd   5.52620         osd.0      up  1.00000 1.00000 
 1   hdd   5.52620         osd.1      up  1.00000 1.00000 
 2   hdd   5.52620         osd.2      up  1.00000 1.00000 
 3   hdd   5.52620         osd.3      up  1.00000 1.00000 
 4   hdd   5.52620         osd.4      up  1.00000 1.00000 
 5   hdd   5.52620         osd.5      up  1.00000 1.00000 
 6   hdd   5.52620         osd.6      up  1.00000 1.00000 
 7   hdd   5.52620         osd.7      up  1.00000 1.00000 
 8   hdd   5.52620         osd.8      up  1.00000 1.00000 
 9   hdd   5.52620         osd.9      up  1.00000 1.00000 
10   hdd   5.52620         osd.10     up  1.00000 1.00000 
11   hdd   5.52620         osd.11     up  1.00000 1.00000 
12   hdd   5.52620         osd.12     up  1.00000 1.00000 
13   hdd   5.52620         osd.13     up  1.00000 1.00000 
14   hdd   5.52620         osd.14     up  1.00000 1.00000 
15   hdd   5.52620         osd.15     up  1.00000 1.00000 
16   hdd   5.52620         osd.16     up  1.00000 1.00000 
17   hdd   5.52620         osd.17     up  1.00000 1.00000 
18   hdd   5.52620         osd.18     up  1.00000 1.00000 
19   hdd   5.52620         osd.19     up  1.00000 1.00000 
20   hdd   5.52620         osd.20     up  1.00000 1.00000 
21   hdd   5.52620         osd.21     up  1.00000 1.00000 
22   hdd   5.52620         osd.22     up  1.00000 1.00000 
23   hdd   5.52620         osd.23     up  1.00000 1.00000 

一共两台服务器,48 个 OSD。需要把 osd8 移除,那么就需要把上面的所有的 24 个 OSD 全部删除。

2. 单个 OSD 进程删除流程

以移除 osd.0 为例看一下移除 OSD 的流程:

2.1 将状态设置成 out

首先要现将 OSD 状态设置成 out。

$ ceph osd out 0
marked out osd.0. 

这个阶段 ceph 会自动将处于 out 状态 OSD 中的数据迁移到其他状态正常的 OSD 上,所以在执行完成后,需要使用 ceph -w 查看数据迁移流程。等到不再有输出后,数据迁移完毕。

$ ceph -w  
  cluster:
    id:     063ed8d6-fc89-4fcb-8811-ff23915983e7
    health: HEALTH_ERR
            12408/606262 objects misplaced (2.047%)
            6 scrub errors
            Reduced data availability: 2 pgs peering
            Possible data damage: 5 pgs inconsistent
            application not enabled on 7 pool(s)
 
  services:
    mon: 3 daemons, quorum dell1,dell2,dell3
    mgr: dell1(active)
    mds: cephfs-1/1/1 up  {0=dell1=up:active}, 2 up:standby
    osd: 48 osds: 48 up, 47 in; 44 remapped pgs
    rgw: 3 daemons active
 
  data:
    pools:   22 pools, 1816 pgs
    objects: 296k objects, 963 GB
    usage:   5222 GB used, 254 TB / 259 TB avail
    pgs:     0.220% pgs not active
             12408/606262 objects misplaced (2.047%)
             1763 active+clean
             29   active+remapped+backfill_wait
             14   active+remapped+backfilling
             5    active+clean+inconsistent
             3    peering
             1    active+recovery_wait
             1    activating+remapped
 
  io:
    client:   59450 kB/s rd, 4419 MB/s wr, 1095 op/s rd, 2848 op/s wr
    recovery: 253 MB/s, 210 keys/s, 123 objects/s
 

2018-07-05 14:21:07.867104 mon.dell1 [WRN] Health check failed: Degraded data redundancy: 7/605732 objects degraded (0.001%), 1 pg degraded (PG_DEGRADED)
2018-07-05 14:21:12.252395 mon.dell1 [INF] Health check cleared: PG_DEGRADED (was: Degraded data redundancy: 7/605732 objects degraded (0.001%), 1 pg degraded)
2018-07-05 14:21:13.510741 mon.dell1 [WRN] Health check update: 12269/606262 objects misplaced (2.024%) (OBJECT_MISPLACED)
2018-07-05 14:21:13.510797 mon.dell1 [INF] Health check cleared: PG_AVAILABILITY (was: Reduced data availability: 2 pgs peering)
2018-07-05 14:21:19.488864 mon.dell1 [WRN] Health check update: 11553/606262 objects misplaced (1.906%) (OBJECT_MISPLACED)
2018-07-05 14:21:25.502619 mon.dell1 [WRN] Health check update: 10504/606262 objects misplaced (1.733%) (OBJECT_MISPLACED)
2018-07-05 14:21:31.745600 mon.dell1 [WRN] Health check update: 10091/606262 objects misplaced (1.664%) (OBJECT_MISPLACED)
2018-07-05 14:21:36.779666 mon.dell1 [WRN] Health check update: 9309/606262 objects misplaced (1.535%) (OBJECT_MISPLACED)
2018-07-05 14:21:41.779947 mon.dell1 [WRN] Health check update: 8580/606262 objects misplaced (1.415%) (OBJECT_MISPLACED)
2018-07-05 14:21:46.816584 mon.dell1 [WRN] Health check update: 8215/606262 objects misplaced (1.355%) (OBJECT_MISPLACED)
2018-07-05 14:21:51.817014 mon.dell1 [WRN] Health check update: 7331/606262 objects misplaced (1.209%) (OBJECT_MISPLACED)
2018-07-05 14:21:56.817406 mon.dell1 [WRN] Health check update: 6929/606262 objects misplaced (1.143%) (OBJECT_MISPLACED)
2018-07-05 14:22:01.817820 mon.dell1 [WRN] Health check update: 6426/606262 objects misplaced (1.060%) (OBJECT_MISPLACED)
2018-07-05 14:22:06.818188 mon.dell1 [WRN] Health check update: 5787/606262 objects misplaced (0.955%) (OBJECT_MISPLACED)
2018-07-05 14:22:11.818606 mon.dell1 [WRN] Health check update: 5429/606262 objects misplaced (0.895%) (OBJECT_MISPLACED)
2018-07-05 14:22:16.818981 mon.dell1 [WRN] Health check update: 5165/606262 objects misplaced (0.852%) (OBJECT_MISPLACED)
2018-07-05 14:22:20.303513 osd.35 [ERR] 13.2ad missing primary copy of 13:b56abc11:::d9593962-fa39-406f-bc35-7e4fcac1be9f.44307.2__shadow_121_1530008810116503747%2fplatform-cms.rar.2~vf7AEOlGNYM1ggI6IhV-iu22oDDXcvS.5_1:head, will try copies on 0
2018-07-05 14:22:21.819353 mon.dell1 [WRN] Health check update: 4866/606262 objects misplaced (0.803%) (OBJECT_MISPLACED)
2018-07-05 14:22:26.819657 mon.dell1 [WRN] Health check update: 4586/606262 objects misplaced (0.756%) (OBJECT_MISPLACED)
2018-07-05 14:22:31.819983 mon.dell1 [WRN] Health check update: 4323/606262 objects misplaced (0.713%) (OBJECT_MISPLACED)
2018-07-05 14:22:36.820335 mon.dell1 [WRN] Health check update: 4113/606262 objects misplaced (0.678%) (OBJECT_MISPLACED)
2018-07-05 14:22:41.820676 mon.dell1 [WRN] Health check update: 3949/606262 objects misplaced (0.651%) (OBJECT_MISPLACED)
2018-07-05 14:22:46.821040 mon.dell1 [WRN] Health check update: 3788/606262 objects misplaced (0.625%) (OBJECT_MISPLACED)
2018-07-05 14:22:51.821395 mon.dell1 [WRN] Health check update: 3665/606262 objects misplaced (0.605%) (OBJECT_MISPLACED)
2018-07-05 14:22:56.821692 mon.dell1 [WRN] Health check update: 3440/606262 objects misplaced (0.567%) (OBJECT_MISPLACED)
2018-07-05 14:23:01.821999 mon.dell1 [WRN] Health check update: 3170/606266 objects misplaced (0.523%) (OBJECT_MISPLACED)
2018-07-05 14:23:06.822355 mon.dell1 [WRN] Health check update: 2956/606266 objects misplaced (0.488%) (OBJECT_MISPLACED)
2018-07-05 14:23:11.822752 mon.dell1 [WRN] Health check update: 2747/606270 objects misplaced (0.453%) (OBJECT_MISPLACED)
2018-07-05 14:23:16.823168 mon.dell1 [WRN] Health check update: 2615/606270 objects misplaced (0.431%) (OBJECT_MISPLACED)
2018-07-05 14:23:21.823523 mon.dell1 [WRN] Health check update: 2512/606270 objects misplaced (0.414%) (OBJECT_MISPLACED)
2018-07-05 14:23:26.823878 mon.dell1 [WRN] Health check update: 2409/606270 objects misplaced (0.397%) (OBJECT_MISPLACED)
2018-07-05 14:23:31.824214 mon.dell1 [WRN] Health check update: 2299/606270 objects misplaced (0.379%) (OBJECT_MISPLACED)
2018-07-05 14:23:36.824596 mon.dell1 [WRN] Health check update: 2194/606270 objects misplaced (0.362%) (OBJECT_MISPLACED)
2018-07-05 14:23:41.825037 mon.dell1 [WRN] Health check update: 2101/606270 objects misplaced (0.347%) (OBJECT_MISPLACED)
2018-07-05 14:23:46.825390 mon.dell1 [WRN] Health check update: 1939/606270 objects misplaced (0.320%) (OBJECT_MISPLACED)
2018-07-05 14:23:51.825725 mon.dell1 [WRN] Health check update: 1777/606270 objects misplaced (0.293%) (OBJECT_MISPLACED)
2018-07-05 14:23:56.826087 mon.dell1 [WRN] Health check update: 1612/606270 objects misplaced (0.266%) (OBJECT_MISPLACED)
2018-07-05 14:24:01.826439 mon.dell1 [WRN] Health check update: 1444/606270 objects misplaced (0.238%) (OBJECT_MISPLACED)
2018-07-05 14:24:06.826755 mon.dell1 [WRN] Health check update: 1315/606270 objects misplaced (0.217%) (OBJECT_MISPLACED)
2018-07-05 14:24:11.828343 mon.dell1 [WRN] Health check update: 1264/606270 objects misplaced (0.208%) (OBJECT_MISPLACED)
2018-07-05 14:24:16.828638 mon.dell1 [WRN] Health check update: 1214/606270 objects misplaced (0.200%) (OBJECT_MISPLACED)
2018-07-05 14:24:21.886644 mon.dell1 [WRN] Health check update: 1161/606270 objects misplaced (0.191%) (OBJECT_MISPLACED)
2018-07-05 14:24:26.887027 mon.dell1 [WRN] Health check update: 1110/606270 objects misplaced (0.183%) (OBJECT_MISPLACED)
2018-07-05 14:24:32.287725 mon.dell1 [WRN] Health check update: 1069/606270 objects misplaced (0.176%) (OBJECT_MISPLACED)
2018-07-05 14:24:39.839578 mon.dell1 [WRN] Health check update: 960/606270 objects misplaced (0.158%) (OBJECT_MISPLACED)
2018-07-05 14:24:45.851276 mon.dell1 [WRN] Health check update: 905/606272 objects misplaced (0.149%) (OBJECT_MISPLACED)
2018-07-05 14:24:51.911053 mon.dell1 [WRN] Health check update: 849/606272 objects misplaced (0.140%) (OBJECT_MISPLACED)
2018-07-05 14:24:57.960803 mon.dell1 [WRN] Health check update: 784/606272 objects misplaced (0.129%) (OBJECT_MISPLACED)
2018-07-05 14:25:05.887641 mon.dell1 [WRN] Health check update: 688/606272 objects misplaced (0.113%) (OBJECT_MISPLACED)
2018-07-05 14:25:11.945922 mon.dell1 [WRN] Health check update: 631/606272 objects misplaced (0.104%) (OBJECT_MISPLACED)
2018-07-05 14:25:16.946267 mon.dell1 [WRN] Health check update: 570/606272 objects misplaced (0.094%) (OBJECT_MISPLACED)
2018-07-05 14:25:21.993994 mon.dell1 [WRN] Health check update: 528/606272 objects misplaced (0.087%) (OBJECT_MISPLACED)
2018-07-05 14:25:26.994417 mon.dell1 [WRN] Health check update: 468/606272 objects misplaced (0.077%) (OBJECT_MISPLACED)
2018-07-05 14:25:31.994789 mon.dell1 [WRN] Health check update: 411/606272 objects misplaced (0.068%) (OBJECT_MISPLACED)
2018-07-05 14:25:36.995192 mon.dell1 [WRN] Health check update: 353/606272 objects misplaced (0.058%) (OBJECT_MISPLACED)
2018-07-05 14:25:42.009567 mon.dell1 [WRN] Health check update: 293/606272 objects misplaced (0.048%) (OBJECT_MISPLACED)
2018-07-05 14:25:47.009879 mon.dell1 [WRN] Health check update: 241/606272 objects misplaced (0.040%) (OBJECT_MISPLACED)
2018-07-05 14:25:52.010822 mon.dell1 [WRN] Health check update: 187/606272 objects misplaced (0.031%) (OBJECT_MISPLACED)
2018-07-05 14:25:57.011182 mon.dell1 [WRN] Health check update: 133/606272 objects misplaced (0.022%) (OBJECT_MISPLACED)
2018-07-05 14:26:02.035637 mon.dell1 [WRN] Health check update: 78/606272 objects misplaced (0.013%) (OBJECT_MISPLACED)
2018-07-05 14:26:07.035965 mon.dell1 [WRN] Health check update: 22/606272 objects misplaced (0.004%) (OBJECT_MISPLACED)
2018-07-05 14:26:12.011546 mon.dell1 [INF] Health check cleared: OBJECT_MISPLACED (was: 22/606272 objects misplaced (0.004%))

2.2 PG 修复

但不是数据迁移结束后就万事大吉了,可以通过下面这个命令看到,数据迁移后,有五个 pg 状态不正常,需要修复。

$ ceph health detail
HEALTH_ERR 6 scrub errors; Possible data damage: 5 pgs inconsistent
OSD_SCRUB_ERRORS 6 scrub errors
PG_DAMAGED Possible data damage: 5 pgs inconsistent
    pg 13.cd is active+clean+inconsistent, acting [20,35]
    pg 13.244 is active+clean+inconsistent, acting [35,22]
    pg 13.270 is active+clean+inconsistent, acting [35,14]
    pg 13.308 is active+clean+inconsistent, acting [35,17]
    pg 13.34f is active+clean+inconsistent, acting [11,35]

执行 repair 命令来修复,如果还是不成功,可以使用 scrub 来进行数据清理。

$ ceph pg repair 13.cd
$ ceph pg scrub 13.cd

2.3 关闭 OSD 进程

数据迁移至此算是完成了,但 osd 进程还是跑着的。

 0   hdd   5.52620         osd.0      up        0 1.00000

接下来需要登录到 OSD 服务器上关闭掉该进程。

$ ssh osd8
$ sudo systemctl stop ceph-osd@0

现在 osd 进程的状态已经已经是 down 了。

 0   hdd   5.52620         osd.0    down        0 1.00000 

2.4 删除 OSD

最后执行 purge 命令,将该 osd 从 CRUSH map 中彻底删掉,至此,单个 OSD 的删除终于完成了。

$ ceph osd purge 0 --yes-i-really-mean-it
purged osd.0

对了,最后,如果 /etc/ceph/ceph.conf 中由对应的该 osd 的信息,记得要一起删除。

3. 参考文档

目录
相关文章
|
运维 Kubernetes Cloud Native
腾讯云私有云平台运维面试
根据会议将面试问题进行总结,很多问题感觉当时没回答好,这是为啥呢?应该还是不熟练吧,或者不善于表达。将次经历分享出来,大家多练练。
859 0
hi3531的pcie atu资源重映射
<p>1. 设置ATU 区域号寄存器为需要配置的地址转换区编号。<br> 2. 设置ATU Region Lower Base Address Register 和ATU Region Upper Base Address<br> Register。(在此区域内的目标地址将由区域号寄存器所在的ATU 转换)<br> 3. 设置ATU Region Limit Address Regi
2915 0
|
存储 Java 编译器
Java泛型类型擦除以及类型擦除带来的问题
Java的泛型采用类型擦除机制,编译后的字节码中泛型信息被清除,仅保留原始类型。例如,`ArrayList&lt;String&gt;`与`ArrayList&lt;Integer&gt;`在运行时被视为相同的`ArrayList`类型。类型擦除导致一些问题: 1. **反射调用泛型方法**:直接调用受限于类型,但通过反射可绕过限制。 2. **类型检查**:编译器先检查泛型类型再擦除,类型检查针对引用而非对象。 3. **自动类型转换**:访问泛型成员时自动插入强制类型转换。
|
JavaScript
使用layui checkbox复选框样式实现单选功能
使用layui checkbox复选框样式实现单选功能
738 0
|
算法 开发工具 git
【git 实用指南】git 增加 本地代码 git add 相关命令和复杂情况需求
【git 实用指南】git 增加 本地代码 git add 相关命令和复杂情况需求
1892 0
|
存储 编译器 数据处理
C语言中二维数组的初始化
C语言中二维数组的初始化
709 0
|
JSON 安全 API
【专栏】四种REST API身份验证方法:基本认证、OAuth、JSON Web Token(JWT)和API密钥
【4月更文挑战第28天】本文探讨了四种REST API身份验证方法:基本认证、OAuth、JSON Web Token(JWT)和API密钥。基本认证简单但不安全;OAuth适用于授权第三方应用;JWT提供安全的身份验证信息传递;API密钥适合内部使用。选择方法时需平衡安全性、用户体验和开发复杂性。
1776 0
|
Oracle 关系型数据库 流计算
flink cdc 同步问题之报错org.apache.flink.util.SerializedThrowable:如何解决
Flink CDC(Change Data Capture)是一个基于Apache Flink的实时数据变更捕获库,用于实现数据库的实时同步和变更流的处理;在本汇总中,我们组织了关于Flink CDC产品在实践中用户经常提出的问题及其解答,目的是辅助用户更好地理解和应用这一技术,优化实时数据处理流程。
|
传感器 监控 安全
基于物联网的智能厨房安全监测系统-下位机程序
基于物联网的智能厨房安全监测系统-下位机程序
326 20
|
关系型数据库 MySQL Java
Mybatis-Plus使用案例(包括初始化以及常用插件)
Mybatis-Plus使用案例(包括初始化以及常用插件)
381 0