I have a cluster where for some OSD the weight-set is defined, while for other OSDs it is not [*].
The OSDs with weight-set defined are Filestore OSDs created years ago using "ceph-disk prepare"
The OSDs where the weight set is not defined are Bluestore OSDs installed recently using
ceph-volume
I think this problem explains while I am not able to use anymore the balancer (after having added these new OSDs).
Any "clean" procedure to fix this mess is appreciated !
Thanks, Massimo
[*]
[root@c-mon-01 /]# ceph osd crush tree
ID CLASS WEIGHT (compat) TYPE NAME
-1 1.95190 root default
-2 0 0 rack Rack15-PianoAlto
-3 0.39038 0.38974 host c-osd-1
0 hdd 0.09760 0.09988 osd.0
1 hdd 0.09760 0.10040 osd.1
2 hdd 0.09760 0.08945 osd.2
3 hdd 0.09760 0.10001 osd.3
-4 0.39038 0.39076 host c-osd-2
4 hdd 0.09760 0.10275 osd.4
5 hdd 0.09760 0.10081 osd.5
6 hdd 0.09760 0.10135 osd.6
7 hdd 0.09760 0.08585 osd.7
-5 0.39038 0.39055 host c-osd-3
8 hdd 0.09760 0.10622 osd.8
9 hdd 0.09760 0.09148 osd.9
10 hdd 0.09760 0.10164 osd.10
11 hdd 0.09760 0.09122 osd.11
-11 0.78076 0.78076 host c-osd-5
12 hdd 0.09760 osd.12
13 hdd 0.09760 osd.13
14 hdd 0.09760 osd.14
15 hdd 0.09760 osd.15
16 hdd 0.09760 osd.16
17 hdd 0.09760 osd.17
18 hdd 0.09760 osd.18
19 hdd 0.09760 osd.19
[root@c-mon-01 /]#
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
