Hello, As part of the python-crush project, I am working on a feature to calculate the available usable space in the pools of a cluster. The idea is to make an accurate and conservative estimate that takes into account the exact PG mappings as well as any other information that could help quantify the variance of data placement ahead of time. To my knowledge, this is not what the current "ceph df" command does.
The theoretical study of the problem is halfway through and I am hoping to get a working POC of that first part soon. Here is link to the said document: http://libcrush.org/xvillaneau/crush-docs/raw/v0.1.0/converted/Ceph%20pool%20capacity%20analysis.pdf Any comment, correction or review is welcome. Additionally, if there are other common pool usage scenarios that could be covered, I will gladly add them in. Best Regards, -- Xavier Villaneau Won't solve clusters filling up too fast, but at least you'll know.
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
