mds, client: only send the metrices supported by MDSes

For the old ceph clusters the clients won't send any metrics to
them as default unless they have backported this commit, but there
has one option 'client_collect_and_send_global_metrics' still could
be used to enable it manually.

This will fix the crash bug when upgrading from old ceph clusters,
which will crash the MDSes once they receive unknown metrics.

Fixes: https://tracker.ceph.com/issues/54411
Signed-off-by: Xiubo Li <xiubli@redhat.com>
(cherry picked from commit e9a26c551c763f75a403ff26f6304d5c10f2ca38)

 Conflicts:
	src/client/Client.cc

diff --git a/src/client/Client.cc b/src/client/Client.cc
index 32f2fd1cf5c0a4ef0ddf23560f0272252fbca855..34d85a4beccf4d119211ce9484c9fd848d603d17 100644 (file)
--- a/src/client/Client.cc
+++ b/src/client/Client.cc
@@ -2221,6 +2221,7 @@ void Client::handle_client_session(const MConstRef<MClientSession>& m)
        break;
       }
       session->mds_features = std::move(m->supported_features);
+      session->mds_metric_flags = std::move(m->metric_spec.metric_flags);
 
       renew_caps(session);
       session->state = MetaSession::STATE_OPEN;
@@ -6711,57 +6712,79 @@ void Client::collect_and_send_global_metrics() {
   std::vector<ClientMetricMessage> message;
 
   // read latency
-  metric = ClientMetricMessage(ReadLatencyPayload(logger->tget(l_c_read)));
-  message.push_back(metric);
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_READ_LATENCY)) {
+    metric = ClientMetricMessage(ReadLatencyPayload(logger->tget(l_c_read)));
+    message.push_back(metric);
+  }
 
   // write latency
-  metric = ClientMetricMessage(WriteLatencyPayload(logger->tget(l_c_wrlat)));
-  message.push_back(metric);
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_WRITE_LATENCY)) {
+    metric = ClientMetricMessage(WriteLatencyPayload(logger->tget(l_c_wrlat)));
+    message.push_back(metric);
+  }
 
   // metadata latency
-  metric = ClientMetricMessage(MetadataLatencyPayload(logger->tget(l_c_lat)));
-  message.push_back(metric);
+  if (session->mds_metric_flags.test(CLIENT_METRIC_TYPE_METADATA_LATENCY)) {
+    metric = ClientMetricMessage(MetadataLatencyPayload(logger->tget(l_c_lat)));
+    message.push_back(metric);
+  }
 
   // cap hit ratio -- nr_caps is unused right now
-  auto [cap_hits, cap_misses] = get_cap_hit_rates();
-  metric = ClientMetricMessage(CapInfoPayload(cap_hits, cap_misses, 0));
-  message.push_back(metric);
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_CAP_INFO)) {
+    auto [cap_hits, cap_misses] = get_cap_hit_rates();
+    metric = ClientMetricMessage(CapInfoPayload(cap_hits, cap_misses, 0));
+    message.push_back(metric);
+  }
 
   // dentry lease hit ratio
-  auto [dlease_hits, dlease_misses, nr] = get_dlease_hit_rates();
-  metric = ClientMetricMessage(DentryLeasePayload(dlease_hits, dlease_misses, nr));
-  message.push_back(metric);
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_DENTRY_LEASE)) {
+    auto [dlease_hits, dlease_misses, nr] = get_dlease_hit_rates();
+    metric = ClientMetricMessage(DentryLeasePayload(dlease_hits, dlease_misses, nr));
+    message.push_back(metric);
+  }
 
   // opened files
-  {
+  if (session->mds_metric_flags.test(CLIENT_METRIC_TYPE_OPENED_FILES)) {
     auto [opened_files, total_inodes] = get_opened_files_rates();
     metric = ClientMetricMessage(OpenedFilesPayload(opened_files, total_inodes));
+    message.push_back(metric);
   }
-  message.push_back(metric);
 
   // pinned i_caps
-  {
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_PINNED_ICAPS)) {
     auto [pinned_icaps, total_inodes] = get_pinned_icaps_rates();
     metric = ClientMetricMessage(PinnedIcapsPayload(pinned_icaps, total_inodes));
+    message.push_back(metric);
   }
-  message.push_back(metric);
 
   // opened inodes
-  {
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_OPENED_INODES)) {
     auto [opened_inodes, total_inodes] = get_opened_inodes_rates();
     metric = ClientMetricMessage(OpenedInodesPayload(opened_inodes, total_inodes));
+    message.push_back(metric);
   }
-  message.push_back(metric);
 
   // read io sizes
-  metric = ClientMetricMessage(ReadIoSizesPayload(total_read_ops,
-                                                  total_read_size));
-  message.push_back(metric);
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_READ_IO_SIZES)) {
+    metric = ClientMetricMessage(ReadIoSizesPayload(total_read_ops,
+                                                    total_read_size));
+    message.push_back(metric);
+  }
 
   // write io sizes
-  metric = ClientMetricMessage(WriteIoSizesPayload(total_write_ops,
-                                                   total_write_size));
-  message.push_back(metric);
+  if (_collect_and_send_global_metrics ||
+      session->mds_metric_flags.test(CLIENT_METRIC_TYPE_WRITE_IO_SIZES)) {
+    metric = ClientMetricMessage(WriteIoSizesPayload(total_write_ops,
+                                                     total_write_size));
+    message.push_back(metric);
+  }
 
   session->con->send_message2(make_message<MClientMetrics>(std::move(message)));
 }

diff --git a/src/client/MetaSession.h b/src/client/MetaSession.h
index c28cd0b83b6e786725bc90d05121d57b3a443a41..d9a1e481aca2d5ae8b0241d96e5af782e52a8aa6 100644 (file)
--- a/src/client/MetaSession.h
+++ b/src/client/MetaSession.h
@@ -25,6 +25,7 @@ struct MetaSession {
   uint64_t cap_renew_seq = 0;
   entity_addrvec_t addrs;
   feature_bitset_t mds_features;
+  feature_bitset_t mds_metric_flags;
 
   enum {
     STATE_NEW, // Unused

diff --git a/src/mds/Server.cc b/src/mds/Server.cc
index 09c024099870ee77e0037c7718b948905587d10d..4fbe058e3063308603771e61d223bb689b6213f5 100644 (file)
--- a/src/mds/Server.cc
+++ b/src/mds/Server.cc
@@ -264,6 +264,7 @@ Server::Server(MDSRank *m, MetricsHandler *metrics_handler) :
   caps_throttle_retry_request_timeout = g_conf().get_val<double>("mds_cap_acquisition_throttle_retry_request_timeout");
   dir_max_entries = g_conf().get_val<uint64_t>("mds_dir_max_entries");
   supported_features = feature_bitset_t(CEPHFS_FEATURES_MDS_SUPPORTED);
+  supported_metric_spec = feature_bitset_t(CEPHFS_METRIC_FEATURES_ALL);
 }
 
 void Server::dispatch(const cref_t<Message> &m)
@@ -855,8 +856,10 @@ void Server::_session_logged(Session *session, uint64_t state_seq, bool open, ve
     metrics_handler->add_session(session);
     ceph_assert(session->get_connection());
     auto reply = make_message<MClientSession>(CEPH_SESSION_OPEN);
-    if (session->info.has_feature(CEPHFS_FEATURE_MIMIC))
+    if (session->info.has_feature(CEPHFS_FEATURE_MIMIC)) {
       reply->supported_features = supported_features;
+      reply->metric_spec = supported_metric_spec;
+    }
     mds->send_message_client(reply, session);
     if (mdcache->is_readonly()) {
       auto m = make_message<MClientSession>(CEPH_SESSION_FORCE_RO);
@@ -1009,8 +1012,10 @@ void Server::finish_force_open_sessions(const map<client_t,pair<Session*,uint64_
         metrics_handler->add_session(session);
 
        auto reply = make_message<MClientSession>(CEPH_SESSION_OPEN);
-       if (session->info.has_feature(CEPHFS_FEATURE_MIMIC))
+       if (session->info.has_feature(CEPHFS_FEATURE_MIMIC)) {
          reply->supported_features = supported_features;
+          reply->metric_spec = supported_metric_spec;
+       }
        mds->send_message_client(reply, session);
 
        if (mdcache->is_readonly())
@@ -1472,8 +1477,10 @@ void Server::handle_client_reconnect(const cref_t<MClientReconnect> &m)
     metrics_handler->add_session(session);
     // notify client of success with an OPEN
     auto reply = make_message<MClientSession>(CEPH_SESSION_OPEN);
-    if (session->info.has_feature(CEPHFS_FEATURE_MIMIC))
+    if (session->info.has_feature(CEPHFS_FEATURE_MIMIC)) {
       reply->supported_features = supported_features;
+      reply->metric_spec = supported_metric_spec;
+    }
     mds->send_message_client(reply, session);
     mds->clog->debug() << "reconnect by " << session->info.inst << " after " << delay;
   }

diff --git a/src/mds/Server.h b/src/mds/Server.h
index 15b5934a727811c7fe7392f0f09147f9dedd65bf..66e9d2136e97e4292052e63761ffb64149ed1470 100644 (file)
--- a/src/mds/Server.h
+++ b/src/mds/Server.h
@@ -487,6 +487,7 @@ private:
   set<client_t> client_reconnect_denied;  // clients whose reconnect msg have been denied .
 
   feature_bitset_t supported_features;
+  feature_bitset_t supported_metric_spec;
   feature_bitset_t required_client_features;
 
   bool forward_all_requests_to_auth = false;