When storing files, Ceph will cut files into several objects while most of which have the same size except the last one(the file size may not be a multiple of the object size).
I know from the official website of Ceph that the object size is not fixed but can be determined by a certain parameter, so which is the parameter deciding the object size?
The Ceph file system is built on top of the RADOS object storage system. That is, the bytes in a file are stored in one or more objects. The way that bytes in a file map onto objects is referred to as the striping strategy. For example, in a striping strategy with fixed 1 MB objects, the bytes in a file at offsets [0, 2**20) may be stored in object.0, and the second megabyte [2**20, 2**21) stored in object.1, and so on. There are other ways to map the linear byte stream onto objects. There is more information here:
http://ceph.com/docs/master/cephfs/file-layouts/
and you can change the striping strategy programmatically when you are using libcephfs:
https://github.com/ceph/ceph/blob/master/src/include/cephfs/libcephfs.h#L738
Notice here that this mapping implies a fixed object size. However, RADOS itself imposes no restrictions on object size (except a configurable maximum size). So if you are using the RADOS object API to store data, there isn't a fixed size.