Reading the thread "s3 requires twice the space it should use", Boris pointed out that the fragmentation for the osds is around 0.8-0.9:
On Thu, Apr 15, 2021 at 8:06 PM Boris Behrens <bb@kervyn.de> wrote:
I also checked the fragmentation on the bluestore OSDs and it is around 0.80 - 0.89 on most OSDs. yikes. [root@s3db1 ~]# ceph daemon osd.23 bluestore allocator score block { "fragmentation_rating": 0.85906054329923576 }
And that made me wonder what is the current recommended (and not recommended) way to handle and reduce the fragmentation of the existing OSDs. Reading around I would think of tweaking the min_alloc_size_{ssd,hdd} and redeploying those OSDs, but I was unable to find much else, I wonder what do people do? ps. There was another thread that got no replies asking something similar (and a bunch of other things): https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/message/3PITWZRNX7R...