From: Patrick Donnelly <pdonnell@redhat.com>
Date: Mon, 16 Sep 2019 00:52:29 +0000 (-0700)
Subject: mds: drive cap recall while dropping cache
X-Git-Tag: v15.1.0~1524^2
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=81e451e0c323d82123a9feb9489816cdedafbf4b;p=ceph.git

mds: drive cap recall while dropping cache

STEADY does not imply TRIM. We want to always trim as many caps as
possible.

Introduced-by: be49866a150e04438cef971dec7948eaf34852fc
Fixes: https://tracker.ceph.com/issues/41835
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
---

diff --git a/src/mds/MDSRank.cc b/src/mds/MDSRank.cc
index 77696da76fc1..4fa74fa87aec 100644
--- a/src/mds/MDSRank.cc
+++ b/src/mds/MDSRank.cc
@@ -329,7 +329,8 @@ private:
     auto duration = std::chrono::duration<double>(now-recall_start).count();
 
     MDSGatherBuilder *gather = new MDSGatherBuilder(g_ceph_context);
-    auto [throttled, count] = server->recall_client_state(gather, Server::RecallFlags::STEADY);
+    auto flags = Server::RecallFlags::STEADY|Server::RecallFlags::TRIM;
+    auto [throttled, count] = server->recall_client_state(gather, flags);
     dout(10) << __func__
              << (throttled ? " (throttled)" : "")
              << " recalled " << count << " caps" << dendl;