On Mon, Nov 20, 2017 at 8:19 AM, Aleksei Gutikov <aleksey.gutikov@xxxxxxxxxx> wrote: > Hello, > > We tried to store large amount of small objects (about 5KB, at least bigger > than 4k and smaller than 8k) in rgw that utilize ec pool > xxx.rgw.buckets.data on ssds. > > osd_pool_erasure_code_stripe_unit=4096 > And we expected that data+xattrs will occupy for each object in that case > max 12k. Maybe some fragmentation occurs, but we did not delete anything. > > Actually we faced space occupied by each object (from "MAX AVAIL" field from > ceph df) is about 64k. > Is that expected? > What configuration option (maybe in bluestore) can improve this, > bluestore_min_alloc_size_ssd? What is your EC config on the pool? What is the output you're using to judge how much space they're taking up? If you're using hard drives, yes, I expect it to require 64KB at minimum anyway...but I would have expected the different shards to take up that space as well. (Though it might just hide in rocksdb and not get counted the same way.) -Greg -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
