понедельник, 24 апреля 2017 г.

Upgrade ceph kraken to luminous in place

1. Remove old ceph-release from all nodes

[ceph1][DEBUG ] Preparing...                          ########################################
[ceph1][WARNIN]         file /etc/yum.repos.d/ceph.repo from install of ceph-release-1-1.el7.noarch conflicts with file from package ceph-release-1-1.el7.noarch

yum remove ceph-release-1-1.el7.noarch

2. Install software on one ceph1 node

ceph-deploy install --release luminous  ceph1

[ceph1][DEBUG ] Complete!
[ceph1][INFO  ] Running command: ceph --version
[ceph1][DEBUG ] ceph version 12.0.2 (5a1b6b3269da99a18984c138c23935e5eb96f73e)
3. Restart services on ceph1

systemctl restart ceph-mon.target
systemctl restart ceph-mgr.target

4. Install software on other nodes ceph2, ceph3

ceph-deploy install --release luminous  ceph2 ceph3

and restart services

systemctl restart ceph-mon.target
systemctl restart ceph-mgr.target

5. Check ceph status

     health HEALTH_OK
     monmap e3: 3 mons at {ceph1=10.1.16.1:6789/0,ceph2=10.1.16.2:6789/0,ceph3=10.1.16.3:6789/0}
            election epoch 96, quorum 0,1,2 ceph1,ceph2,ceph3
      fsmap e27: 1/1/1 up {0=ceph1=up:active}
        mgr active: ceph1 standbys: ceph3, ceph2
     osdmap e4649: 12 osds: 12 up, 12 in
      pgmap v226087: 1608 pgs, 4 pools, 2765 GB data, 691k objects
            5578 GB used, 5151 GB / 10729 GB avail
                1608 active+clean
  client io 10384 B/s wr, 0 op/s rd, 2 op/s wr

On each node restart OSD consistently, wait until the apopriate OSD's are up

systemctl restart ceph-osd.target


[root@ceph1 ceph]# ceph osd tree
ID WEIGHT   TYPE NAME      UP/DOWN REWEIGHT PRIMARY-AFFINITY
-1 10.47839 root default
-2  3.49280     host ceph1
 0  0.87320         osd.0       up  1.00000          1.00000
 3  0.87320         osd.3       up  1.00000          1.00000
 6  0.87320         osd.6       up  1.00000          1.00000
 9  0.87320         osd.9       up  1.00000          1.00000
-3  3.49280     host ceph2
 1  0.87320         osd.1       up  1.00000          1.00000
 4  0.87320         osd.4       up  1.00000          1.00000
 7  0.87320         osd.7       up  1.00000          1.00000
10  0.87320         osd.10      up  1.00000          1.00000
-4  3.49280     host ceph3
12  0.87320         osd.12      up  1.00000          1.00000
13  0.87320         osd.13      up  1.00000          1.00000
14  0.87320         osd.14      up  1.00000          1.00000
15  0.87320         osd.15      up  1.00000          1.00000

and wait health ok

6. Set osd flag
I have the warning:

health HEALTH_WARN
all OSDs are running luminous or later but the 'require_luminous_osds' osdmap flag is not set

Execute:
ceph osd set require_luminous_osds


7. Restart other services

systemctl restart ceph-mds.target

8. Finally check status
[root@ceph1 ceph]# ceph -s
    cluster 2d856a9e-ff9e-4322-bedb-56f3be26b607
     health HEALTH_OK
     monmap e3: 3 mons at {ceph1=10.1.16.1:6789/0,ceph2=10.1.16.2:6789/0,ceph3=10.1.16.3:6789/0}
            election epoch 96, quorum 0,1,2 ceph1,ceph2,ceph3
      fsmap e33: 1/1/1 up {0=ceph1=up:active}
        mgr active: ceph1 standbys: ceph3, ceph2
     osdmap e4674: 12 osds: 12 up, 12 in
      pgmap v227520: 1608 pgs, 4 pools, 2765 GB data, 691k objects
            5573 GB used, 5156 GB / 10729 GB avail
                1608 active+clean
  client io 17394 B/s wr, 0 op/s rd, 2 op/s wr

Комментариев нет:

Отправить комментарий