On Tue, 11 Nov 2014 17:14:49 +0530 Mallikarjun Biradar wrote: > Hi all > > When Issued ceph osd dump it displays weight for that osd as 1 and when > issued osd tree it displays 0.35 > There are many threads about this, google is your friend. For example: https://www.mail-archive.com/ceph-users@xxxxxxxxxxxxxx/msg11010.html In short, one is the CRUSH weight (usually based on the capacity of the OSD), the other is the OSD weight (or reweight in the tree display). For example think about a cluster with 100 2TB OSDs and you're planning to replace them (bit by bit) with 4TB OSDs. But the hard disks are the same speed, so if you would just replace things, more and more data would migrate to your bigger OSDs, making the whole cluster actually slower. Setting the OSD weight (reweight) to 0.5 for the 4TB OSDs (untiil the replacement is complete) will result in them getting the same allocation as the 2TB ones, keeping things even. Christian > output from osd dump: > { "osd": 20, > "uuid": "b2a97a29-1b8a-43e4-a4b0-fd9ee351086e", > "up": 1, > "in": 1, > "weight": "1.000000", > "primary_affinity": "1.000000", > "last_clean_begin": 0, > "last_clean_end": 0, > "up_from": 103, > "up_thru": 106, > "down_at": 0, > "lost_at": 0, > "public_addr": "10.242.43.116:6820\/27623", > "cluster_addr": "10.242.43.116:6821\/27623", > "heartbeat_back_addr": "10.242.43.116:6822\/27623", > "heartbeat_front_addr": "10.242.43.116:6823\/27623", > "state": [ > "exists", > "up"]}], > > output from osd tree: > # id weight type name up/down reweight > -1 7.35 root default > -2 2.8 host rack6-storage-5 > 0 0.35 osd.0 up 1 > 1 0.35 osd.1 up 1 > 2 0.35 osd.2 up 1 > 3 0.35 osd.3 up 1 > 4 0.35 osd.4 up 1 > 5 0.35 osd.5 up 1 > 6 0.35 osd.6 up 1 > 7 0.35 osd.7 up 1 > -3 2.8 host rack6-storage-4 > 8 0.35 osd.8 up 1 > 9 0.35 osd.9 up 1 > 10 0.35 osd.10 up 1 > 11 0.35 osd.11 up 1 > 12 0.35 osd.12 up 1 > 13 0.35 osd.13 up 1 > 14 0.35 osd.14 up 1 > 15 0.35 osd.15 up 1 > -4 1.75 host rack6-storage-6 > 16 0.35 osd.16 up 1 > 17 0.35 osd.17 up 1 > 18 0.35 osd.18 up 1 > 19 0.35 osd.19 up 1 > 20 0.35 osd.20 up 1 > > Please help me to understand this > > -regards, > Mallikarjun Biradar -- Christian Balzer Network/Systems Engineer chibi@xxxxxxx Global OnLine Japan/Fusion Communications http://www.gol.com/ _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |