From: xie xingguo <xie.xingguo@zte.com.cn>
Date: Thu, 13 Jul 2017 00:21:48 +0000 (+0800)
Subject: osd/OSD: gracefully shutdown on error exit during init
X-Git-Tag: ses5-milestone9~1^2~13^2~4
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=a98c7e84ac12d3e6fb4076cf52cab59c4aa9d59e;p=ceph.git

osd/OSD: gracefully shutdown on error exit during init

This can avoid crashes as below:

  0> 2017-07-12 09:34:47.427438 7f320ce61b80 -1 /home/xxg/build/ceph-dev/src/common/HeartbeatMap.cc: In function 'ceph::HeartbeatMap::~HeartbeatMap()'
thread 7f320ce61b80 time 2017-07-12 09:34:47.422986
/home/xxg/build/ceph-dev/src/common/HeartbeatMap.cc: 39: FAILED assert(m_workers.empty())

 ceph version 12.1.0-702-gc5b99af (c5b99af) luminous (rc)
 1: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x110) [0x7f320d8ba7f0]
 2: (ceph::HeartbeatMap::~HeartbeatMap()+0xf8) [0x7f320d9be0a8]
 3: (CephContext::~CephContext()+0x40c) [0x7f320d9a648c]
 4: (CephContext::put()+0xe6) [0x7f320d9a6776]
 5: (main()+0xad3) [0x7f320d282953]
 6: (__libc_start_main()+0xf5) [0x7f32094cfb15]
 7: (()+0x4964c9) [0x7f320d31f4c9]
 NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
---

diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index 0e0b6ba25ec6..eeb5bc5d8480 100644
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -2568,9 +2568,11 @@ int OSD::init()
 
   r = monc->authenticate();
   if (r < 0) {
+    derr << __func__ << " authentication failed: " << cpp_strerror(r)
+         << dendl;
     osd_lock.Lock(); // locker is going to unlock this on function exit
     if (is_stopping())
-      r =  0;
+      r = 0;
     goto monout;
   }
 
@@ -2578,9 +2580,10 @@ int OSD::init()
     derr << "unable to obtain rotating service keys; retrying" << dendl;
     ++rotating_auth_attempts;
     if (rotating_auth_attempts > g_conf->max_rotating_auth_attempts) {
+        derr << __func__ << " wait_auth_rotating timed out" << dendl;
         osd_lock.Lock(); // make locker happy
         if (!is_stopping()) {
-            r = - ETIMEDOUT;
+            r = -ETIMEDOUT;
         }
         goto monout;
     }
@@ -2588,12 +2591,16 @@ int OSD::init()
 
   r = update_crush_device_class();
   if (r < 0) {
+    derr << __func__ <<" unable to update_crush_device_class: "
+         << cpp_strerror(r) << dendl;
     osd_lock.Lock();
     goto monout;
   }
 
   r = update_crush_location();
   if (r < 0) {
+    derr << __func__ <<" unable to update_crush_location: "
+         << cpp_strerror(r) << dendl;
     osd_lock.Lock();
     goto monout;
   }
@@ -2629,8 +2636,7 @@ int OSD::init()
 
   return 0;
 monout:
-  mgrc.shutdown();
-  monc->shutdown();
+  exit(1);
 
 out:
   enable_disable_fuse(true);