mon/OSDMonitor: respect CRUSH weights for reweight-by-pg

Do not assume that all OSDs are weighted equally for reweight-by-pg.

Note that reweight-by-utilization already reweights based on the size of
the OSD volume; we presume that this is already reflected by the CRUSH
weights.

Signed-off-by: Sage Weil <sage@redhat.com>

diff --git a/src/mon/OSDMonitor.cc b/src/mon/OSDMonitor.cc
index 7eb90f13be68afc775f0ed7b145a235f9078be9a..1b92f06ea7f730dd9add2ac366faef6e115aec84 100644 (file)
--- a/src/mon/OSDMonitor.cc
+++ b/src/mon/OSDMonitor.cc
@@ -459,13 +459,14 @@ int OSDMonitor::reweight_by_utilization(int oload, std::string& out_str,
   const PGMap &pgm = mon->pgmon()->pg_map;
   vector<int> pgs_by_osd(osdmap.get_max_osd());
   unsigned num_pg_copies = 0;
-  unsigned num_osds = pgm.osd_stat.size();
 
   // Avoid putting a small number (or 0) in the denominator when calculating
   // average_util
   double average_util;
   if (by_pg) {
     // by pg mapping
+    double weight_sum = 0.0;      // sum up the crush weights
+    int num_osds = 0;
     for (ceph::unordered_map<pg_t,pg_stat_t>::const_iterator p =
           pgm.pg_stat.begin();
         p != pgm.pg_stat.end();
@@ -477,6 +478,10 @@ int OSDMonitor::reweight_by_utilization(int oload, std::string& out_str,
           ++q) {
        if (*q >= (int)pgs_by_osd.size())
          pgs_by_osd.resize(*q);
+       if (pgs_by_osd[*q] == 0) {
+         weight_sum += osdmap.crush->get_item_weightf(*q);
+         ++num_osds;
+       }
        ++pgs_by_osd[*q];
        ++num_pg_copies;
       }
@@ -490,7 +495,7 @@ int OSDMonitor::reweight_by_utilization(int oload, std::string& out_str,
       return -EDOM;
     }
 
-    average_util = (double)num_pg_copies / (double)num_osds;
+    average_util = (double)num_pg_copies / weight_sum;
   } else {
     // by osd utilization
     if (pgm.osd_sum.kb < 1024) {
@@ -532,7 +537,7 @@ int OSDMonitor::reweight_by_utilization(int oload, std::string& out_str,
        ++p) {
     float util;
     if (by_pg) {
-      util = pgs_by_osd[p->first];
+      util = pgs_by_osd[p->first] / osdmap.crush->get_item_weightf(p->first);
     } else {
       util = (double)p->second.kb_used / (double)p->second.kb;
     }