From: Samuel Just <sam.just@inktank.com>
Date: Tue, 26 Mar 2013 20:08:29 +0000 (-0700)
Subject: PG: make _select_auth_object smarter
X-Git-Tag: v0.62~113^2~9
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=e61c94e2f6943731438621cec71aec84012b46ab;p=ceph.git

PG: make _select_auth_object smarter

Previously, we just picked the first one to have the object in
question.  Now, we will attempt to choose one that has as
much of the following as possible:
1) has the object (there must be one)
2) has an object_info attr
3) has a valid object_info attr
4) has an object_info whose size matches the scrubbed size

Signed-off-by: Samuel Just <sam.just@inktank.com>
---

diff --git a/src/osd/PG.cc b/src/osd/PG.cc
index c3208b3f44e..85edc07f4fa 100644
--- a/src/osd/PG.cc
+++ b/src/osd/PG.cc
@@ -4239,17 +4239,61 @@ map<int, ScrubMap *>::const_iterator PG::_select_auth_object(
   const map<int,ScrubMap*> &maps)
 {
   map<int, ScrubMap *>::const_iterator auth = maps.end();
-  // Select first present replica
   for (map<int, ScrubMap *>::const_iterator j = maps.begin();
        j != maps.end();
        ++j) {
-    if (!j->second->objects.count(obj)) {
+    map<hobject_t, ScrubMap::object>::iterator i =
+      j->second->objects.find(obj);
+    if (i == j->second->objects.end()) {
       continue;
     }
     if (auth == maps.end()) {
+      // Something is better than nothing
+      // TODO: something is NOT better than nothing, do something like
+      // unfound_lost if no valid copies can be found, or just mark unfound
       auth = j;
+      dout(10) << __func__ << ": selecting osd " << j->first
+	       << " for obj " << obj
+	       << ", auth == maps.end()"
+	       << dendl;
+      continue;
+    }
+    map<string, bufferptr>::iterator k = i->second.attrs.find(OI_ATTR);
+    if (k == i->second.attrs.end()) {
+      // no object info on object, probably corrupt
+      dout(10) << __func__ << ": rejecting osd " << j->first
+	       << " for obj " << obj
+	       << ", no oi attr"
+	       << dendl;
+      continue;
+    }
+    bufferlist bl;
+    bl.push_back(k->second);
+    object_info_t oi;
+    try {
+      bufferlist::iterator bliter = bl.begin();
+      ::decode(oi, bliter);
+    } catch (...) {
+      dout(10) << __func__ << ": rejecting osd " << j->first
+	       << " for obj " << obj
+	       << ", corrupt oi attr"
+	       << dendl;
+      // invalid object info, probably corrupt
+      continue;
+    }
+    if (oi.size != i->second.size) {
+      // invalid size, probably corrupt
+      dout(10) << __func__ << ": rejecting osd " << j->first
+	       << " for obj " << obj
+	       << ", size mismatch"
+	       << dendl;
+      // invalid object info, probably corrupt
       continue;
     }
+    dout(10) << __func__ << ": selecting osd " << j->first
+	     << " for obj " << obj
+	     << dendl;
+    auth = j;
   }
   return auth;
 }