From: Dan van der Ster Date: Wed, 18 Jun 2025 19:28:16 +0000 (-0700) Subject: common/options: add option osd_crush_scaling_factor X-Git-Url: http://git.apps.os.sepia.ceph.com/?a=commitdiff_plain;h=e890bde763a329405145bf6d5ebf8b2d3c74a2b7;p=ceph.git common/options: add option osd_crush_scaling_factor Signed-off-by: Dan van der Ster --- diff --git a/src/common/options/global.yaml.in b/src/common/options/global.yaml.in index 0999c791ba67f..634bdc326433f 100644 --- a/src/common/options/global.yaml.in +++ b/src/common/options/global.yaml.in @@ -6834,3 +6834,21 @@ options: level: dev default: 0 desc: When EC writes should generate PDWs (development only) 0=optimal 1=never 2=when possible +- name: osd_crush_scaling_factor + type: float + level: advanced + desc: Scale device sizes by this factor when calculating CRUSH weights. + This value is applied when new OSDs are added to the cluster, + or when running 'ceph osd crush reweight-by-scaling-factor'. + E.g. When set to 1.0, CRUSH weights will be the same as the device size in TiB and + the maximum cluster size is ~65.5PiB. + If set to 0.1, CRUSH weights will be 1/10th of the device size in TiB and + the maximum cluster size is ~655PiB. + default: 1.0 + min: 0.001 + max: 1.0 + services: + - mon + - osd + flags: + - runtime