osd: fine-grained statistics of logical object space usage

To test this change, we create an image of 5GB and do rbd bench write of 1GB:
./bin/rbd create bar -s 5120 && ./bin/rbd bench --io-type write --io-size 32K --io-total 100M --io-pattern rand  rbd/bar

Below is the test result.

Was:

GLOBAL:
    SIZE       AVAIL      RAW USED     %RAW USED
    30911M     27052M        3859M         12.49
POOLS:
    NAME                  ID     USED      %USED     MAX AVAIL     OBJECTS
    rbd                   0      3191M     26.36         8914M        1174
    cephfs_data_a         1          0         0         8914M           0
    cephfs_metadata_a     2       2246         0         8914M          21

Now:

GLOBAL:
    SIZE       AVAIL      RAW USED     %RAW USED
    30911M     27050M        3861M         12.49
POOLS:
    NAME                  ID     USED        %USED     MAX AVAIL     OBJECTS
    rbd                   0      101216k      1.10         8913M        1178
    cephfs_data_a         1            0         0         8913M           0
    cephfs_metadata_a     2          892         0         8913M          21

E.g., this change can make "osd pool set-quota max_bytes" work nicely.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>

src/osd/PrimaryLogPG.cc

@@ -6893,10 +6893,15 @@ void PrimaryLogPG::write_update_size_and_usage(object_stat_sum_t& delta_stats, o
   modified.union_of(ch);
   if (write_full || offset + length > oi.size) {
     uint64_t new_size = offset + length;
-    delta_stats.num_bytes -= oi.size;
-    delta_stats.num_bytes += new_size;
     oi.size = new_size;
   }
+  { // count newly write bytes, exclude overwrites
+    interval_set<uint64_t> overlap;
+    overlap.intersection_of(ch, oi.extents);
+    ch.subtract(overlap);
+    oi.extents.union_of(ch);
+    delta_stats.num_bytes += ch.size();
+  }
   delta_stats.num_wr++;
   delta_stats.num_wr_kb += SHIFT_ROUND_UP(length, 10);
 }

src/osd/osd_types.cc

@@ -4967,7 +4967,7 @@ void object_info_t::encode(bufferlist& bl, uint64_t features) const
        ++i) {
     old_watchers.insert(make_pair(i->first.second, i->second));
   }
-  ENCODE_START(16, 8, bl);
+  ENCODE_START(17, 8, bl);
   ::encode(soid, bl);
   ::encode(myoloc, bl);	//Retained for compatibility
   ::encode((__u32)0, bl); // was category, no longer used
@@ -4998,13 +4998,14 @@ void object_info_t::encode(bufferlist& bl, uint64_t features) const
   ::encode(expected_object_size, bl);
   ::encode(expected_write_size, bl);
   ::encode(alloc_hint_flags, bl);
+  ::encode(extents, bl);
   ENCODE_FINISH(bl);
 }
 
 void object_info_t::decode(bufferlist::iterator& bl)
 {
   object_locator_t myoloc;
-  DECODE_START_LEGACY_COMPAT_LEN(16, 8, 8, bl);
+  DECODE_START_LEGACY_COMPAT_LEN(17, 8, 8, bl);
   map<entity_name_t, watch_info_t> old_watchers;
   ::decode(soid, bl);
   ::decode(myoloc, bl);
@@ -5086,6 +5087,9 @@ void object_info_t::decode(bufferlist::iterator& bl)
     expected_write_size = 0;
     alloc_hint_flags = 0;
   }
+  if (struct_v >= 17) {
+    ::decode(extents, bl);
+  }
   DECODE_FINISH(bl);
 }
 

src/osd/osd_types.h

@@ -4479,6 +4479,7 @@ struct object_info_t {
   // alloc hint attribute
   uint64_t expected_object_size, expected_write_size;
   uint32_t alloc_hint_flags;
+  interval_set<uint64_t> extents; // deduplicated logic extent map
 
   void copy_user_bits(const object_info_t& other);