From: Sage Weil <sage@redhat.com>
Date: Tue, 31 Jan 2017 20:14:59 +0000 (-0500)
Subject: osd: do not send ENXIO on misdirected op by default
X-Git-Tag: v12.0.0~19^2
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=refs%2Fpull%2F13206%2Fhead;p=ceph.git

osd: do not send ENXIO on misdirected op by default

In practice this tends to get bubbled up the stack as an error on
the caller, and they usually do not handle it properly.  For example,
with librbd, this turns into EIO and break the VM.

Instead, this will manifest as a hung op on the client.  That is
also not ideal, but given that the root cause here is generally a
bug, it's not clear what else would be better.

We already log an error in the cluster log, so teuthology runs will
continue to fail.

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

diff --git a/PendingReleaseNotes b/PendingReleaseNotes
index 65347cc2b26..99a1d9e3b7d 100644
--- a/PendingReleaseNotes
+++ b/PendingReleaseNotes
@@ -30,3 +30,11 @@
    either the IoCtx methods on older librados versions or the
    deprecated methods on any version of librados will lead to
    incomplete results if/when the new OSD limits are enabled.
+
+* In previous versions, if a client sent an op to the wrong OSD, the OSD
+  would reply with ENXIO.  The rationale here is that the client or OSD is
+  clearly buggy and we want to surface the error as clearly as possible.
+  We now only send the ENXIO reply if the osd_enxio_on_misdirected_op option
+  is enabled (it's off by default).  This means that a VM using librbd that
+  previously would have gotten an EIO and gone read-only will now see a
+  blocked/hung IO instead.
diff --git a/src/common/config_opts.h b/src/common/config_opts.h
index ff8125ef480..8529fd0c018 100644
--- a/src/common/config_opts.h
+++ b/src/common/config_opts.h
@@ -853,6 +853,7 @@ OPTION(osd_debug_reject_backfill_probability, OPT_DOUBLE, 0)
 OPTION(osd_debug_inject_copyfrom_error, OPT_BOOL, false)  // inject failure during copyfrom completion
 OPTION(osd_debug_randomize_hobject_sort_order, OPT_BOOL, false)
 OPTION(osd_debug_misdirected_ops, OPT_BOOL, false)
+OPTION(osd_enxio_on_misdirected_op, OPT_BOOL, false)
 OPTION(osd_debug_verify_cached_snaps, OPT_BOOL, false)
 OPTION(osd_enable_op_tracker, OPT_BOOL, true) // enable/disable OSD op tracking
 OPTION(osd_num_op_tracker_shard, OPT_U32, 32) // The number of shards for holding the ops
diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index 1359c8b2956..577ed5de5e2 100644
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -1434,11 +1434,14 @@ void OSDService::handle_misdirected_op(PG *pg, OpRequestRef op)
 
   dout(7) << *pg << " misdirected op in " << m->get_map_epoch() << dendl;
   clog->warn() << m->get_source_inst() << " misdirected " << m->get_reqid()
-	      << " pg " << m->get_pg()
-	      << " to osd." << whoami
-	      << " not " << pg->acting
-	      << " in e" << m->get_map_epoch() << "/" << osdmap->get_epoch() << "\n";
-  reply_op_error(op, -ENXIO);
+	       << " pg " << m->get_pg()
+	       << " to osd." << whoami
+	       << " not " << pg->acting
+	       << " in e" << m->get_map_epoch() << "/" << osdmap->get_epoch()
+	       << "\n";
+  if (g_conf->osd_enxio_on_misdirected_op) {
+    reply_op_error(op, -ENXIO);
+  }
 }
 
 
@@ -8785,14 +8788,16 @@ void OSD::handle_op(OpRequestRef& op, OSDMapRef& osdmap)
   if (!send_map->osd_is_valid_op_target(pgid.pgid, whoami)) {
     dout(7) << "we are invalid target" << dendl;
     clog->warn() << m->get_source_inst() << " misdirected " << m->get_reqid()
-		      << " pg " << m->get_pg()
-		      << " to osd." << whoami
-		      << " in e" << osdmap->get_epoch()
-		      << ", client e" << m->get_map_epoch()
-		      << " pg " << pgid
-		      << " features " << m->get_connection()->get_features()
-		      << "\n";
-    service.reply_op_error(op, -ENXIO);
+		 << " pg " << m->get_pg()
+		 << " to osd." << whoami
+		 << " in e" << osdmap->get_epoch()
+		 << ", client e" << m->get_map_epoch()
+		 << " pg " << pgid
+		 << " features " << m->get_connection()->get_features()
+		 << "\n";
+    if (g_conf->osd_enxio_on_misdirected_op) {
+      service.reply_op_error(op, -ENXIO);
+    }
     return;
   }