crimson/os/seastore: make segment cleaner do gc based on gc benefit cost ratio

Segments' benefit cost ratio is calculated:
segment_last_mod_time = max{last modification time of all extents in the segment}
segment_last_rewrite_time = max{last rewrite time of all extents in the segment}
segment_utilization = segment_live_bytes / segment_size
segment_age = current_time - max{segment_last_mod_time, segment_last_rewrite_time}
benefit_cost_ratio = (1 - segment_utilization) * segment_age / (1 + segment_utilization)

Segment cleaner always reclaim the segment with the largest benefit cost ratio

Signed-off-by: Xuehan Xu <[email protected]>

Author: xxhdx1985126

src/crimson/os/seastore/segment_cleaner.h

@@ -895,28 +895,41 @@ class SegmentCleaner : public SegmentProvider {
 
   // journal status helpers
 
+  double calc_gc_benefit_cost(segment_id_t id) const {
+    double util = space_tracker->calc_utilization(id);
+    auto cur_time = seastar::lowres_system_clock::now();
+    auto segment = segments[id];
+    assert(cur_time >= segment.last_modified);
+    auto segment_age =
+      cur_time - std::max(segment.last_modified, segment.last_rewritten);
+    uint64_t age = segment_age.count();
+    return (1 - util) * age / (1 + util);
+  }
+
   journal_seq_t get_next_gc_target() const {
     segment_id_t id = NULL_SEG_ID;
     segment_seq_t seq = NULL_SEG_SEQ;
-    int64_t least_live_bytes = std::numeric_limits<int64_t>::max();
+    double max_benefit_cost = 0;
     for (auto it = segments.begin();
 	 it != segments.end();
 	 ++it) {
       auto _id = it->first;
       const auto& segment_info = it->second;
+      double benefit_cost = calc_gc_benefit_cost(_id);
       if (segment_info.is_closed() &&
 	  !segment_info.is_in_journal(journal_tail_committed) &&
-	  space_tracker->get_usage(_id) < least_live_bytes) {
+	  benefit_cost > max_benefit_cost) {
 	id = _id;
 	seq = segment_info.journal_segment_seq;
-	least_live_bytes = space_tracker->get_usage(id);
+	max_benefit_cost = benefit_cost;
       }
     }
     if (id != NULL_SEG_ID) {
       crimson::get_logger(ceph_subsys_seastore_cleaner).debug(
-	"SegmentCleaner::get_next_gc_target: segment {} seq {}",
+	"SegmentCleaner::get_next_gc_target: segment {} seq {}, benefit_cost {}",
 	id,
-	seq);
+	seq,
+	max_benefit_cost);
       return journal_seq_t{seq, paddr_t::make_seg_paddr(id, 0)};
     } else {
       return JOURNAL_SEQ_NULL;