osd: do not join cluster if not healthy

If our internal heartbeats are failing, do not send a boot message and try
to join the cluster.

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

diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index 0e3bade863adf4452d8c27a4f396e6ff150871de..82a9a54e281e01068d56674e615e77874b69afab 100644 (file)
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -2277,6 +2277,14 @@ void OSD::tick()
 
   logger->set(l_osd_buf, buffer::get_total_alloc());
 
+  if (is_waiting_for_healthy()) {
+    if (g_ceph_context->get_heartbeat_map()->is_healthy()) {
+      dout(1) << "healthy again, booting" << dendl;
+      state = STATE_BOOTING;
+      start_boot();
+    }
+  }
+
   if (is_active()) {
     // periodically kick recovery work queue
     recovery_tp.wake();
@@ -2694,6 +2702,13 @@ void OSD::_maybe_boot(epoch_t oldest, epoch_t newest)
     return;
   }
 
+  // if we are not healthy, do not mark ourselves up (yet)
+  if (!g_ceph_context->get_heartbeat_map()->is_healthy()) {
+    dout(5) << "not healthy, deferring boot" << dendl;
+    state = STATE_WAITING_FOR_HEALTHY;
+    return;
+  }
+
   // if our map within recent history, try to add ourselves to the osdmap.
   if (osdmap->test_flag(CEPH_OSDMAP_NOUP)) {
     dout(5) << "osdmap NOUP flag is set, waiting for it to clear" << dendl;

diff --git a/src/osd/OSD.h b/src/osd/OSD.h
index 359d75396fead1e2f3ee5c87ba4c2de951be640b..1fb34aac6108700bc853b68240a53f346b847ded 100644 (file)
--- a/src/osd/OSD.h
+++ b/src/osd/OSD.h
@@ -494,6 +494,7 @@ public:
   static const int STATE_BOOTING = 2;
   static const int STATE_ACTIVE = 3;
   static const int STATE_STOPPING = 4;
+  static const int STATE_WAITING_FOR_HEALTHY = 5;
 
 private:
   int state;
@@ -506,6 +507,7 @@ public:
   bool is_booting() { return state == STATE_BOOTING; }
   bool is_active() { return state == STATE_ACTIVE; }
   bool is_stopping() { return state == STATE_STOPPING; }
+  bool is_waiting_for_healthy() { return state == STATE_WAITING_FOR_HEALTHY; }
 
 private: