From 98af31d10f362c05ea8ed57495973b08599431e7 Mon Sep 17 00:00:00 2001
From: Rishabh Dave <ridave@redhat.com>
Date: Wed, 17 Oct 2018 12:42:43 +0000
Subject: [PATCH] mds: don't mark unresponsive sessions holding no caps stale

When an unresponsive MDS session holds no caps, do not mark it stale
even after session_timeout; at session_autoclose, evict it directly.

Fixes: http://tracker.ceph.com/issues/17854
Signed-off-by: Rishabh Dave <ridave@redhat.com>
---
 src/mds/Locker.cc    |  8 ++++++++
 src/mds/Locker.h     |  1 +
 src/mds/Server.cc    | 15 +++++++++++++++
 src/mds/SessionMap.h |  6 ++++++
 4 files changed, 30 insertions(+)

diff --git a/src/mds/Locker.cc b/src/mds/Locker.cc
index 31d941de892..2d5bfb88501 100644
--- a/src/mds/Locker.cc
+++ b/src/mds/Locker.cc
@@ -2622,6 +2622,14 @@ void Locker::mark_need_snapflush_inode(CInode *in)
   }
 }
 
+bool Locker::is_revoking_any_caps_from(client_t client)
+{
+  auto it = revoking_caps_by_client.find(client);
+  if (it == revoking_caps_by_client.end())
+    return false;
+  return !it->second.empty();
+}
+
 void Locker::_do_null_snapflush(CInode *head_in, client_t client, snapid_t last)
 {
   dout(10) << "_do_null_snapflush client." << client << " on " << *head_in << dendl;
diff --git a/src/mds/Locker.h b/src/mds/Locker.h
index 7ed6e7a1dc6..dd8494816de 100644
--- a/src/mds/Locker.h
+++ b/src/mds/Locker.h
@@ -208,6 +208,7 @@ protected:
 public:
   void snapflush_nudge(CInode *in);
   void mark_need_snapflush_inode(CInode *in);
+  bool is_revoking_any_caps_from(client_t client);
 
   // local
 public:
diff --git a/src/mds/Server.cc b/src/mds/Server.cc
index d4392d1b604..b23dbd89bfd 100644
--- a/src/mds/Server.cc
+++ b/src/mds/Server.cc
@@ -956,6 +956,21 @@ void Server::find_idle_sessions()
 	}
       }
 
+      if (last_cap_renew_span >= mds->mdsmap->get_session_autoclose()) {
+	dout(20) << "evicting session " << session->info.inst << " since autoclose "
+		    "has arrived" << dendl;
+	// evict session without marking it stale
+	to_evict.push_back(session);
+	continue;
+      }
+
+      if (!session->is_any_flush_waiter() &&
+	  !mds->locker->is_revoking_any_caps_from(session->get_client())) {
+	dout(20) << "deferring marking session " << session->info.inst << " stale "
+		    "since it holds no caps" << dendl;
+	continue;
+      }
+
       auto it = session->info.client_metadata.find("timeout");
       if (it != session->info.client_metadata.end()) {
 	unsigned timeout = strtoul(it->second.c_str(), nullptr, 0);
diff --git a/src/mds/SessionMap.h b/src/mds/SessionMap.h
index abcaab679d0..0dfbc51b5ac 100644
--- a/src/mds/SessionMap.h
+++ b/src/mds/SessionMap.h
@@ -300,13 +300,19 @@ public:
   void touch_cap(Capability *cap) {
     caps.push_front(&cap->item_session_caps);
   }
+
   void touch_cap_bottom(Capability *cap) {
     caps.push_back(&cap->item_session_caps);
   }
+
   void touch_lease(ClientLease *r) {
     leases.push_back(&r->item_session_lease);
   }
 
+  bool is_any_flush_waiter() {
+    return !waitfor_flush.empty();
+  }
+
   // -- leases --
   uint32_t lease_seq = 0;
 
-- 
2.39.5