mds: avoid sending duplicated discovers during recovery

If MDS just entered the rejoin state, it should not kick discovers
because the discovers were just sent. Similarly, if MDS just entered
the clientreplay state, it should not call MDS::handle_mds_recovery()
because MDS::recovery_done() has already recovered the table server.

Also make MDCache::handle_mds_recovery() not wake the discover waiters
up. Because the MDCache::kick_discovers re-sends the discovers, their
replies will wake the discover waiter up.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>

diff --git a/src/mds/MDCache.cc b/src/mds/MDCache.cc
index eb20c06085a0c8603bac1e6a3d80dc717b6f427d..167a85e75f8e18e4c6dd9076e8777c655599b411 100644 (file)
--- a/src/mds/MDCache.cc
+++ b/src/mds/MDCache.cc
@@ -2777,6 +2777,10 @@ void MDCache::handle_mds_recovery(int who)
 {
   dout(7) << "handle_mds_recovery mds." << who << dendl;
 
+  // exclude all discover waiters. kick_discovers() will do the job
+  static const uint64_t i_mask = CInode::WAIT_ANY_MASK & ~CInode::WAIT_DIR;
+  static const uint64_t d_mask = CDir::WAIT_ANY_MASK & ~CDir::WAIT_DENTRY;
+
   list<Context*> waiters;
 
   // wake up any waiters in their subtrees
@@ -2797,7 +2801,7 @@ void MDCache::handle_mds_recovery(int who)
     while (!q.empty()) {
       CDir *d = q.front();
       q.pop_front();
-      d->take_waiting(CDir::WAIT_ANY_MASK, waiters);
+      d->take_waiting(d_mask, waiters);
 
       // inode waiters too
       for (CDir::map_t::iterator p = d->items.begin();
@@ -2806,7 +2810,7 @@ void MDCache::handle_mds_recovery(int who)
        CDentry *dn = p->second;
        CDentry::linkage_t *dnl = dn->get_linkage();
        if (dnl->is_primary()) {
-         dnl->get_inode()->take_waiting(CInode::WAIT_ANY_MASK, waiters);
+         dnl->get_inode()->take_waiting(i_mask, waiters);
          
          // recurse?
          list<CDir*> ls;
@@ -9904,8 +9908,11 @@ void MDCache::kick_discovers(int who)
 {
   for (map<tid_t,discover_info_t>::iterator p = discovers.begin();
        p != discovers.end();
-       ++p)
+       ++p) {
+    if (p->second.mds != who)
+      continue;
     _send_discover(p->second);
+  }
 }
 
 

diff --git a/src/mds/MDS.cc b/src/mds/MDS.cc
index b8ae455a60d02a29798902e78e519a3165319c18..a2e7228cbcdae46146a874139daa3937824f6a70 100644 (file)
--- a/src/mds/MDS.cc
+++ b/src/mds/MDS.cc
@@ -1037,28 +1037,31 @@ void MDS::handle_mds_map(MMDSMap *m)
     if (g_conf->mds_dump_cache_after_rejoin &&
        oldmap->is_rejoining() && !mdsmap->is_rejoining()) 
       mdcache->dump_cache();      // for DEBUG only
-  
-    // ACTIVE|CLIENTREPLAY|REJOIN => we can discover from them.
-    set<int> olddis, dis;
-    oldmap->get_mds_set(olddis, MDSMap::STATE_ACTIVE);
-    oldmap->get_mds_set(olddis, MDSMap::STATE_CLIENTREPLAY);
-    oldmap->get_mds_set(olddis, MDSMap::STATE_REJOIN);
-    mdsmap->get_mds_set(dis, MDSMap::STATE_ACTIVE);
-    mdsmap->get_mds_set(dis, MDSMap::STATE_CLIENTREPLAY);
-    mdsmap->get_mds_set(dis, MDSMap::STATE_REJOIN);
-    for (set<int>::iterator p = dis.begin(); p != dis.end(); ++p) 
-      if (*p != whoami &&            // not me
-         olddis.count(*p) == 0) {  // newly so?
-       mdcache->kick_discovers(*p);
-       mdcache->kick_open_ino_peers(*p);
-      }
+
+    if (oldstate >= MDSMap::STATE_REJOIN) {
+      // ACTIVE|CLIENTREPLAY|REJOIN => we can discover from them.
+      set<int> olddis, dis;
+      oldmap->get_mds_set(olddis, MDSMap::STATE_ACTIVE);
+      oldmap->get_mds_set(olddis, MDSMap::STATE_CLIENTREPLAY);
+      oldmap->get_mds_set(olddis, MDSMap::STATE_REJOIN);
+      mdsmap->get_mds_set(dis, MDSMap::STATE_ACTIVE);
+      mdsmap->get_mds_set(dis, MDSMap::STATE_CLIENTREPLAY);
+      mdsmap->get_mds_set(dis, MDSMap::STATE_REJOIN);
+      for (set<int>::iterator p = dis.begin(); p != dis.end(); ++p)
+       if (*p != whoami &&            // not me
+           olddis.count(*p) == 0) {  // newly so?
+         mdcache->kick_discovers(*p);
+         mdcache->kick_open_ino_peers(*p);
+       }
+    }
   }
 
   if (oldmap->is_degraded() && !mdsmap->is_degraded() && state >= MDSMap::STATE_ACTIVE)
     dout(1) << "cluster recovered." << dendl;
 
   // did someone go active?
-  if (is_clientreplay() || is_active() || is_stopping()) {
+  if (oldstate >= MDSMap::STATE_CLIENTREPLAY &&
+      (is_clientreplay() || is_active() || is_stopping())) {
     set<int> oldactive, active;
     oldmap->get_mds_set(oldactive, MDSMap::STATE_ACTIVE);
     oldmap->get_mds_set(oldactive, MDSMap::STATE_CLIENTREPLAY);