On Thu, 8 Jan 2015 15:35:22 -0600 Michael J Brewer wrote: > > > Hi all, > > I'm working on filling a cluster to near capacity for testing purposes. > Though I'm noticing that it isn't storing the data uniformly between OSDs > during the filling process. I currently have the following levels: >
Reading this ML and/or it's archives and thus google are your friend. See for example the current "Different disk usage on different OSDs" thread. [snip] > > I am using "rados put" to upload 100g files to the cluster, doing two at > a time from two different locations. Is this expected behavior, or can > someone shed light on why it is doing this? We're using the opensource > version 80.7. We're also using the default CRUSH configuration. > Assuming that this is test cluster, empty it to at least 50% again, completely if possible to reduce the time for data movement. Verify that you have the optimal amount of PGs (see the current calculator thread) and adjust if needed. Watch the data rejigger itself. Set your crush tunables to optimal. Another round of happy data movement. Once the dust settles, you ought to have a better, but no way perfect distribution. It comes with the territory, Ceph needs to be deterministic first. See the mentioned threads again. Christian -- Christian Balzer Network/Systems Engineer ch...@gol.com Global OnLine Japan/Fusion Communications http://www.gol.com/ _______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
