ceph集群故障运维--持续更新

一.PG处于异常状态active+undersized+degraded

部署环境: 自己搭建的3节点集群，集群共5个OSD，部署Ceph的RadosGW的服务时，副本默认设置为3,集群存放数据量少。集群状态处于如下，无法自己恢复：

[root@k8s-01 ceph]# ceph -s
  cluster:
    id:     dd1a1ab2-0f34-4936-bc09-87bd40ef5ca0
    health: HEALTH_WARN
            Degraded data redundancy: 183/4019 objects degraded (4.553%), 15 pgs degraded, 16 pgs undersized

  services:
    mon: 3 daemons, quorum k8s-01,k8s-02,k8s-03
    mgr: k8s-01(active)
    mds: cephfs-2/2/2 up  {0=k8s-01=up:active,1=k8s-03=up:active}, 1 up:standby
    osd: 5 osds: 5 up, 5 in
    rgw: 3 daemons active

  data:
    pools:   6 pools, 288 pgs
    objects: 1.92 k objects, 1020 MiB
    usage:   7.5 GiB used, 342 GiB / 350 GiB avail
    pgs:     183/4019 objects degraded (4.553%)
             272 active+clean
             15  active+undersized+degraded
             1   active+undersized

分析上面两个状态显示的具体含义:

• undersized
  
  PG当前的Acting Set小于存储池副本数

• degraded
  
  降级状态，Peering完成后，PG检测到任意一个PG实例存在不一致(需要被同步/修复)的对象，或者当前ActingSet小于存储池副本数。

以我们的集群为例说明，这个状态降级的集群可以正常读写数据，undersized是当前存活的PG副本数为2，小于副本数3.将其做此标记，表明存数据副本数不足。

针对我们的集群，我们查看下集群的详细信息：

[root@k8s-01 ceph]# ceph health detail
HEALTH_WARN Degraded data redundancy: 183/4017 objects degraded (4.556%), 15 pgs degraded, 16 pgs undersized
PG_DEGRADED Degraded data redundancy: 183/4017 objects degraded (4.556%), 15 pgs degraded, 16 pgs undersized
    pg 4.0 is stuck undersized for 782.927699, current state active+undersized+degraded, last acting [0,3]
    pg 4.1 is stuck undersized for 782.930776, current state active+undersized+degraded, last acting [1,4]
    pg 4.2 is stuck undersized for 782.930141, current state active+undersized, last acting [1,3]
    pg 4.3 is stuck undersized for 782.929428, current state active+undersized+degraded, last acting [0,4]
    pg 4.4 is stuck undersized for 782.931202, current state active+undersized+degraded, last acting [1,3]
    pg 4.5 is stuck undersized for 782.931627, current state active+undersized+degraded, last acting [3,0]
    pg 4.6 is stuck undersized for 782.931584, current state active+undersized+degraded, last acting [0,4]
    pg 4.7 is stuck undersized for 782.928895, current state active+undersized+degraded, last acting [1,3]
    pg 6.0 is stuck undersized for 766.814237, current state active+undersized+degraded, last acting [0,4]
    pg 6.1 is stuck undersized for 766.818367, current state active+undersized+degraded, last acting [1,3]
    pg 6.2 is stuck undersized for 766.819767, current state active+undersized+degraded, last acting [1,3]
    pg 6.3 is stuck undersized for 766.822230, current state active+undersized+degraded, last acting [4,2]
    pg 6.4 is stuck undersized for 766.821061, current state active+undersized+degraded, last acting [1,4]
    pg 6.5 is stuck undersized for 766.826778, current state active+undersized+degraded, last acting [3,0]
    pg 6.6 is stuck undersized for 766.818134, current state active+undersized+degraded, last acting [1,3]
    pg 6.7 is stuck undersized for 766.826165, current state active+undersized+degraded, last acting [3,2]
[root@k8s-01 ceph]#

分析:这两种状态一般同时出现，表明有些PG没有满足设定的replicas数量要求。由上面的信息我们分析PG 4.0显示只有两个拷贝，分别在OSD.0和OSD.3上，分析其原因可能是我们的集群副本数无法满足。我们的集群主机是以host为故障域，所以我们设置副本数为3时，需要保证每个节点都必须存在OSD，结果查看集群的osd tree发现只有两个节点有OSD，另外一个节点不存在OSD，这里我们设置存储池的副本为2,查看集群是否正常。

# 查看ceph的OSD磁盘大小
[root@k8s-01 ceph]# ceph osd df
ID CLASS WEIGHT  REWEIGHT SIZE    USE     AVAIL   %USE VAR  PGS
 0   hdd 0.09769  1.00000 100 GiB 1.5 GiB  98 GiB 1.50 0.70 118
 1   hdd 0.09769  1.00000 100 GiB 1.4 GiB  99 GiB 1.42 0.66 123
 2   hdd 0.04880  1.00000  50 GiB 1.5 GiB  48 GiB 3.00 1.40  47
 3   hdd 0.04880  1.00000  50 GiB 1.6 GiB  48 GiB 3.14 1.46 151
 4   hdd 0.04880  1.00000  50 GiB 1.5 GiB  48 GiB 3.04 1.42 137
                    TOTAL 350 GiB 7.5 GiB 342 GiB 2.14
MIN/MAX VAR: 0.66/1.46  STDDEV: 0.83
# 查看ceph的OSD tree
[root@k8s-01 ceph]# ceph osd tree
ID CLASS WEIGHT  TYPE NAME       STATUS REWEIGHT PRI-AFF
-1       0.34177 root default
-3       0.24417     host k8s-01
 0   hdd 0.09769         osd.0       up  1.00000 1.00000
 1   hdd 0.09769         osd.1       up  1.00000 1.00000
 2   hdd 0.04880         osd.2       up  1.00000 1.00000
-5       0.09760     host k8s-02
 3   hdd 0.04880         osd.3       up  1.00000 1.00000
 4   hdd 0.04880         osd.4       up  1.00000 1.00000

• 设置存储池的副本数为2

[root@k8s-01 ceph]# ceph osd  pool set default.rgw.log size 2
[root@k8s-01 ceph]# ceph osd dump
epoch 130
fsid dd1a1ab2-0f34-4936-bc09-87bd40ef5ca0
created 2019-06-03 19:22:38.483687
modified 2019-07-08 18:14:50.342952
flags sortbitwise,recovery_deletes,purged_snapdirs
crush_version 11
full_ratio 0.95
backfillfull_ratio 0.9
nearfull_ratio 0.85
require_min_compat_client jewel
min_compat_client jewel
require_osd_release mimic
pool 1 'datapool' replicated size 2 min_size 1 crush_rule 0 object_hash rjenkins pg_num 128 pgp_num 128 last_change 48 flags hashpspool stripe_width 0 application cephfs
pool 2 'metapool' replicated size 2 min_size 1 crush_rule 0 object_hash rjenkins pg_num 128 pgp_num 128 last_change 52 flags hashpspool stripe_width 0 application cephfs
pool 3 '.rgw.root' replicated size 2 min_size 1 crush_rule 0 object_hash rjenkins pg_num 8 pgp_num 8 last_change 121 owner 18446744073709551615 flags hashpspool stripe_width 0 application rgw
pool 4 'default.rgw.control' replicated size 2 min_size 1 crush_rule 0 object_hash rjenkins pg_num 8 pgp_num 8 last_change 127 flags hashpspool stripe_width 0 application rgw
pool 5 'default.rgw.meta' replicated size 2 min_size 1 crush_rule 0 object_hash rjenkins pg_num 8 pgp_num 8 last_change 126 flags hashpspool stripe_width 0 application rgw
pool 6 'default.rgw.log' replicated size 2 min_size 1 crush_rule 0 object_hash rjenkins pg_num 8 pgp_num 8 last_change 129 owner 18446744073709551615 flags hashpspool stripe_width 0 application rgw
max_osd 5

• 查看集群状态

[root@k8s-01 ceph]# ceph -s
  cluster:
    id:     dd1a1ab2-0f34-4936-bc09-87bd40ef5ca0
    health: HEALTH_OK

  services:
    mon: 3 daemons, quorum k8s-01,k8s-02,k8s-03
    mgr: k8s-01(active)
    mds: cephfs-2/2/2 up  {0=k8s-01=up:active,1=k8s-03=up:active}, 1 up:standby
    osd: 5 osds: 5 up, 5 in
    rgw: 3 daemons active

  data:
    pools:   6 pools, 288 pgs
    objects: 1.92 k objects, 1019 MiB
    usage:   7.5 GiB used, 342 GiB / 350 GiB avail
    pgs:     288 active+clean

  io:
    client:   339 B/s rd, 11 KiB/s wr, 0 op/s rd, 1 op/s wr

至此我们发现集群状态已经恢复。

二.cephfs挂载点执行df -h卡死

• 查看卡死的客户端ID

[root@k8s-02 ~]# ceph daemon mds.k8s-02 session ls
{
        "id": 44272,
        "num_leases": 0,
        "num_caps": 1,
        "state": "open",
        "request_load_avg": 0,
        "uptime": 230.958432,
        "replay_requests": 0,
        "completed_requests": 0,
        "reconnecting": false,
        "inst": "client.44272 192.168.50.11:0/2167123153",
        "client_metadata": {
            "features": "00000000000001ff",
            "ceph_sha1": "cbff874f9007f1869bfd3821b7e33b2a6ffd4988",
            "ceph_version": "ceph version 13.2.5 (cbff874f9007f1869bfd3821b7e33b2a6ffd4988) mimic (stable)",
            "entity_id": "admin",
            "hostname": "k8s-02",
            "mount_point": "/mnt/ceph",
            "pid": "1904636",

• 执行evict

# 这里evict后面是客户端的ID
[root@k8s-02 ~]# cep daemon mds.k8s-02 session evict 44272

• umount -l 卡死的挂载点

[root@k8s-02 ~]# umount -l /mnt/ceph
[root@k8s-02 ~]# ceph-fuse /mnt/ceph