2

According to Ceph's documentation, when storing a file into Ceph, it is split into many objects first, and CRUSH will then map each object to a placement group.

When we read a file from Ceph, we only input the key to to get the file, how does Ceph know how many objects the file is split into, and how does it find all objects belong to a file ?

0xF2
  • 314
  • 3
  • 17
Fish sec
  • 21
  • 2
  • Did you find any answer for your question? – Mehran Akhavan Jun 26 '20 at 10:19
  • please cite the documentation precisely, I don't think it splits data by default in a replicated, i.e., non-erasure-coded, pool. Also, you tagged cephfs, but your question does not really make it clear if you mean RADOS (plain objects) or actually files on CephFS – hbogert Aug 14 '20 at 18:52

0 Answers0