Answes inline. 2018-05-25 17:57 GMT+02:00 Jesus Cea <j...@jcea.es>:
> Hi there. > > I have configured a POOL with a 8+2 erasure code. My target by space > usage and OSD configuration, would be 128 PG, but since each configure > PG will be using 10 actual "PGs", I have created the pool with only 8 PG > (80 real PG). Since I can increase PGs but not decreasing it, this > decision seems sensible. > > Some questions: > > 1. Documentation insists everywhere that the PG could should be a power > of two. Would be nice to know the consequences of not following this > recommendation. Would be nice to know too if being "close" to a power of > two is better than be far away and if it is better to be close but below > or close but a little bit more. If ideal value is 128 but I only can be > 120 or 130, what should I choose?. 120 or 130?. Why? > Go for the next larger power of two under the assumption that your cluster will grow. > > 2. As I understand, the PG count that should be "power of two" is "8", > in this case (real 80 PG underneath). Good. In this case, the next step > would be 16 (160 real PG). I would rather prefer to increase it to 12 or > 13 (120/130 real PGs). Would it be reasonable?. What are the > consequences of increasing PG to 12 or 13 instead of choosing 16 (the > next power of two). > Data will be poorly balanced between PGs if it's not a power of two. > > 3. Is there any negative effect for CRUSH of using erasure code 8+2 > instead of 6+2 or 14+2 (power of two)?. I have 25 OSDs, so requiring 16 > for a single operation seems a bad idea, even more when my OSD > capacities are very spread (from 150 GB to 1TB) and filling a small OSD > would block writes in the entire pool. > EC rules don't have to be powers of two. And yes, too many chunks for EC pools is a bad idea. It's rarely advisable to have a total of k + m larger than 8 or so. Also, you should have at least k + m + 1 servers, otherwise full server failures cannot be handled properly. A large spread between the OSD capacities within one crush rule is also usually a bad idea, 150 GB to 1 TB is typically too big. > > 4. Since I have created a erasure coded pool with 8 PG, I am getting > warnings of "x pools have many more objects per pg than average". The > data I am copying is coming from a legacy pool with PG=512. New pool PG > is 8. That is creating ~30.000 objects per PG, far above average (616 > objects). What can I do?. Moving to 16 or 32 PGs is not going to improve > the situation, but will consume PGs (32*10). Advice?. > Well, you reduced the number of PGs by a factor of 64, so you'll of course see a large skew here. The option mon_pg_warn_max_object_skew controls when this warning is shown, default is 10. > > 5. I understand the advice of having <300 PGs per OSD because memory > usage, but I am wondering about the impact of the number of objects in > each PG. I wonder if memory and resource wise, having 100 PG with 10.000 > objects each is far more demanding than 1000 PGs with 50 objects each. > Since I have PGs with 300 objects and PGs with 30.000 objects, I wonder > about the memory impact of each. What is the actual memory hungry factor > in a OSD, PGs or objects per PG?. > PGs typically impose a bigger overhead. But PGs with a large number of objects can become annoying... Paul > > Thanks for your time and knowledge :). > > -- > Jesús Cea Avión _/_/ _/_/_/ _/_/_/ > j...@jcea.es - http://www.jcea.es/ _/_/ _/_/ _/_/ _/_/ _/_/ > Twitter: @jcea _/_/ _/_/ _/_/_/_/_/ > jabber / xmpp:j...@jabber.org _/_/ _/_/ _/_/ _/_/ _/_/ > "Things are not so easy" _/_/ _/_/ _/_/ _/_/ _/_/ _/_/ > "My name is Dump, Core Dump" _/_/_/ _/_/_/ _/_/ _/_/ > "El amor es poner tu felicidad en la felicidad de otro" - Leibniz > > > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > > -- Paul Emmerich Looking for help with your Ceph cluster? Contact us at https://croit.io croit GmbH Freseniusstr. 31h 81247 München www.croit.io Tel: +49 89 1896585 90
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com