mds/MDBalancer: separate out fragmentation from hit_dir

In _fragment_finish we would like to try splitting
the new frags without applying a spurious hit to their
temperatures.  Same for the start of openc where we would
like to do an early check for splitting without hitting
the dir twice.

Signed-off-by: John Spray <john.spray@redhat.com>

diff --git a/src/mds/MDBalancer.cc b/src/mds/MDBalancer.cc
index 334c11e94b537161e51f5e95051394645971a029..5f6d279fbe4961cbf889b5c895fa65d58c205589 100644 (file)
--- a/src/mds/MDBalancer.cc
+++ b/src/mds/MDBalancer.cc
@@ -1061,36 +1061,25 @@ void MDBalancer::hit_inode(utime_t now, CInode *in, int type, int who)
     hit_dir(now, in->get_parent_dn()->get_dir(), type, who);
 }
 
-void MDBalancer::hit_dir(utime_t now, CDir *dir, int type, int who, double amount)
+void MDBalancer::maybe_fragment(CDir *dir, bool hot)
 {
-  // hit me
-  double v = dir->pop_me.get(type).hit(now, amount);
-
   // split/merge
   if (g_conf->mds_bal_frag && g_conf->mds_bal_fragment_interval > 0 &&
       !dir->inode->is_base() &&        // not root/base (for now at least)
       dir->is_auth()) {
 
-    dout(20) << "hit_dir " << type << " pop is " << v << ", frag " << dir->get_frag()
-            << " size " << dir->get_frag_size() << dendl;
-
     // split
     if (g_conf->mds_bal_split_size > 0 &&
        mds->mdsmap->allows_dirfrags() &&
-       (dir->should_split() ||
-        (v > g_conf->mds_bal_split_rd && type == META_POP_IRD) ||
-        (v > g_conf->mds_bal_split_wr && type == META_POP_IWR))) {
-      dout(10) << "hit_dir " << type << " pop is " << v
-               << ", putting in split_pending: " << *dir << dendl;
+       (dir->should_split() || hot))
+    {
       if (split_pending.count(dir->dirfrag()) == 0) {
         queue_split(dir, false);
       } else {
         if (dir->should_split_fast()) {
-          dout(4) << "hit_dir: fragment hit hard limit, splitting immediately ("
-            << *dir << ")" << dendl;
           queue_split(dir, true);
         } else {
-          dout(10) << "hit_dir: fragment already enqueued to split: "
+          dout(10) << __func__ << ": fragment already enqueued to split: "
                    << *dir << dendl;
         }
       }
@@ -1099,10 +1088,23 @@ void MDBalancer::hit_dir(utime_t now, CDir *dir, int type, int who, double amoun
     // merge?
     if (dir->get_frag() != frag_t() && dir->should_merge() &&
        merge_pending.count(dir->dirfrag()) == 0) {
-      dout(10) << "hit_dir " << type << " pop is " << v << ", putting in merge_pending: " << *dir << dendl;
       queue_merge(dir);
     }
   }
+}
+
+void MDBalancer::hit_dir(utime_t now, CDir *dir, int type, int who, double amount)
+{
+  // hit me
+  double v = dir->pop_me.get(type).hit(now, amount);
+
+  const bool hot = (v > g_conf->mds_bal_split_rd && type == META_POP_IRD) ||
+                   (v > g_conf->mds_bal_split_wr && type == META_POP_IWR);
+
+  dout(20) << "hit_dir " << type << " pop is " << v << ", frag " << dir->get_frag()
+           << " size " << dir->get_frag_size() << dendl;
+
+  maybe_fragment(dir, hot);
 
   // replicate?
   if (type == META_POP_IRD && who >= 0) {

diff --git a/src/mds/MDBalancer.h b/src/mds/MDBalancer.h
index 3a53db2a13e983802aae9fd0c929916976365f6b..438ffdeb8375521524dca2c06a84c57824fde658 100644 (file)
--- a/src/mds/MDBalancer.h
+++ b/src/mds/MDBalancer.h
@@ -124,6 +124,14 @@ public:
 
   void queue_split(const CDir *dir, bool fast);
   void queue_merge(CDir *dir);
+
+  /**
+   * Based on size and configuration, decide whether to issue a queue_split
+   * or queue_merge for this CDir.
+   *
+   * \param hot whether the directory's temperature is enough to split it
+   */
+  void maybe_fragment(CDir *dir, bool hot);
 };