Regarding Primary affinity configuration

["Johnu George (johnugeo)" <johnugeo@xxxxxxxxx>]

Hi All,

          I have few questions regarding the Primary affinity.  In the original blueprint (https://wiki.ceph.com/Planning/Blueprints/Firefly/osdmap%3A_primary_role_affinity ), one example has been given.  

For PG x, CRUSH returns [a, b, c]

If a has primary_affinity of .5, b and c have 1 , with 50% probability, we will choose b or c instead of a. (25% for b, 25% for c)

A) I was browsing through the code, but I could not find this logic of splitting the rest of configured primary affinity value between other osds. How is this handled?

1.     if (a < CEPH_OSD_MAX_PRIMARY_AFFINITY &&
2.     (crush_hash32_2(CRUSH_HASH_RJENKINS1,
3.             seed, o) >> 16) >= a) {
4.       // we chose not to use this primary.  note it anyway as a
5.       // fallback in case we don't pick anyone else, but keep looking.
6.       if (pos < 0)
7.     pos = i;
8.     } else {
9.       pos = i;
10.       break;
11.     }
12.   }

B) Since, primary affinity value is configured independently, there can be a situation with [0.1,0.1,0.1]  with total value that don’t add to 1.  How is this taken care of?  

C) Slightly confused. What happens for a situation with [1,0.5,1] ? Is osd.0 always returned?

D) After calculating primary based on the affinity values, I see a shift of osds so that primary comes to the front. Why is this needed?. I thought, primary affinity value affects only reads and hence, osd ordering need not be changed. 

Thanks,

Johnu

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com