Hoststatus remove stats creation. (#8689)

* Update traffic_ctl to query HostStatus directly

- Updates HostStatus to eliminate stats creation.
- Updates traffic_ctl JSON-RPC to query HostStatus directly.
- adds in a new persistance mechanism for HosStatus.

* Changes from review comments.

* remove redundant 'const'.

Author: jrushford

doc/appendices/command-line/traffic_ctl_jsonrpc.en.rst

@@ -325,11 +325,10 @@ traffic_ctl host
 ----------------
 .. program:: traffic_ctl host
 
-A stat to track status is created for each host. The name is the host fqdn with a prefix of
-"proxy.process.host_status". The value of the stat is a string which is the serialized
-representation of the status. This contains the overall status and the status for each reason.  The
-stats may be viewed using the :program:`traffic_ctl metric` command or through the `stats_over_http`
-endpoint.
+A record to track status is created for each host. The name is the host fqdn.  The value of the
+record when retrieved, is a serialized string representation of the status.
+This contains the overall status and the status for each reason.  The
+records may be viewed using the :program:`traffic_ctl host status` command.
 
 .. option:: --time count
 
@@ -366,7 +365,7 @@ endpoint.
    :ref:`admin_lookup_records`
 
    Get the current status of the specified hosts with respect to their use as targets for parent
-   selection. This returns the same information as the per host stat.
+   selection. If the HOSTNAME arguments are omitted, all host records available are returned.
 
 .. option:: down HOSTNAME [HOSTNAME ...]
 

doc/developer-guide/jsonrpc/jsonrpc-api.en.rst

@@ -1106,9 +1106,9 @@ admin_host_set_status
 Description
 ~~~~~~~~~~~
 
-A stat to track status is created for each host. The name is the host fqdn with a prefix of `proxy.process.host_status`. The value of
-the stat is a string which is the serialized representation of the status. This contains the overall status and the status for each reason.
-The stats may be viewed using the `admin_lookup_records`_ rpc api or through the ``stats_over_http`` endpoint.
+A record to track status is created for each host. The name is the host fqdn.
+This record contains the overall status and the status for each reason.
+The records may be viewed using the `admin_host_get_status` rpc api.
 
 Parameters
 ~~~~~~~~~~
@@ -1204,11 +1204,8 @@ Response:
 Getting the host status
 ~~~~~~~~~~~~~~~~~~~~~~~
 
-Get the current status of the specified hosts with respect to their use as targets for parent selection. This returns the same
-information as the per host stat.
-
-Although there is no specialized API that you can call to get a status from a particular host you can work away by pulling the right records.
-For instance, the ``host1``  that we just set up can be easily fetch for a status:
+Get the current status of the specified hosts with respect to their use as targets for parent selection. This returns the serialized
+information for the host.
 
 Request:
 
@@ -1218,11 +1215,10 @@ Request:
    {
       "id": "ded7018e-0720-11eb-abe2-001fc69cc946",
       "jsonrpc": "2.0",
-      "method": "admin_lookup_records",
-      "params": [{
-            "record_name": "proxy.process.host_status.host1"
-         }
-      ]
+      "method": "admin_host_get_status",
+      "params": [
+            "host1.mycdn.net"
+       ]
    }
 
 Response:
@@ -1234,23 +1230,11 @@ Response:
       "jsonrpc": "2.0",
       "id": "ded7018e-0720-11eb-abe2-001fc69cc946",
       "result": {
-         "recordList": [{
-            "record": {
-               "record_name": "proxy.process.host_status.host1",
-               "record_type": "3",
-               "version": "0",
-               "raw_stat_block": "0",
-               "order": "1134",
-               "stat_meta": {
-                  "persist_type": "1"
-               },
-               "record_class": "2",
-               "overridable": "false",
-               "data_type": "STRING",
-               "current_value": "HOST_STATUS_UP,ACTIVE:UP:0:0,LOCAL:UP:0:0,MANUAL:UP:0:0,SELF_DETECT:UP:0",
-               "default_value": "HOST_STATUS_UP,ACTIVE:UP:0:0,LOCAL:UP:0:0,MANUAL:UP:0:0,SELF_DETECT:UP:0"
+         "statusList": [{
+            "hostname": "host1.mycdn.net",
+            "status": "HOST_STATUS_DOWN,ACTIVE:UP:0:0,LOCAL:UP:0:0,MANUAL:UP:0:0,SELF_DETECT:DOWN:1646248306"
             }
-         }]
+         ]
          ,"errorList":[]
       }
    }

include/shared/rpc/yaml_codecs.h

@@ -240,4 +240,4 @@ class yamlcpp_json_emitter
 
     return resp;
   }
-};
+};

include/tscore/Filenames.h

@@ -46,6 +46,7 @@ namespace filename
   ///////////////////////////////////////////////////////////////////
   // Various other file names
   constexpr const char *RECORDS_STATS = "records.snap";
+  constexpr const char *HOST_RECORDS  = "host_records.yaml";
 
 } // namespace filename
 } // namespace ts

proxy/HostStatus.h

@@ -33,6 +33,8 @@
 #include <ctime>
 #include <string>
 #include <sstream>
+#include "tscore/Filenames.h"
+#include "tscore/I_Layout.h"
 #include "tscore/ink_rwlock.h"
 #include "records/P_RecProcess.h"
 
@@ -89,6 +91,11 @@ struct Reason {
   }
 };
 
+struct HostStatuses {
+  std::string hostname;
+  std::string status;
+};
+
 // host status POD
 struct HostStatRec {
   TSHostStatus status;
@@ -168,6 +175,33 @@ struct HostStatRec {
   }
 };
 
+struct HostStatusSync : public Continuation {
+  std::string hostRecordsFile;
+
+  void sync_task();
+  void
+  getHostStatusPersistentFilePath()
+  {
+    std::string rundir(RecConfigReadRuntimeDir());
+    hostRecordsFile = Layout::relative_to(rundir, ts::filename::HOST_RECORDS);
+  }
+
+public:
+  HostStatusSync()
+  {
+    getHostStatusPersistentFilePath();
+
+    SET_HANDLER(&HostStatusSync::mainEvent);
+  }
+
+  int
+  mainEvent(int event, Event *e)
+  {
+    sync_task();
+    return EVENT_DONE;
+  }
+};
+
 /**
  * Singleton placeholder for next hop status.
  */
@@ -182,11 +216,16 @@ struct HostStatus {
   }
   void setHostStatus(const std::string_view name, const TSHostStatus status, const unsigned int down_time,
                      const unsigned int reason);
-  HostStatRec *getHostStatus(const std::string_view name);
-  void createHostStat(const std::string_view name, const char *data = nullptr);
-  void loadHostStatusFromStats();
+  void loadFromPersistentStore();
   void loadRecord(std::string_view name, HostStatRec &h);
-  RecErrT getHostStat(std::string &stat_name, char *buf, unsigned int buf_len);
+  HostStatRec *getHostStatus(const std::string_view name);
+  void getAllHostStatuses(std::vector<HostStatuses> &hosts);
+  std::string
+  getHostStatusPersistentFilePath()
+  {
+    std::string rundir(RecConfigReadRuntimeDir());
+    return Layout::relative_to(rundir, ts::filename::HOST_RECORDS);
+  }
 
 private:
   HostStatus();

src/traffic_ctl_jsonrpc/CtrlCommands.cc

@@ -312,11 +312,8 @@ void
 HostCommand::status_get()
 {
   auto const &data = _arguments.get("status");
-  HostGetStatusRequest request;
-  for (auto it : data) {
-    std::string name = std::string{HostGetStatusRequest::STATUS_PREFIX} + "." + it;
-    request.emplace_rec(name, shared::rpc::NOT_REGEX, shared::rpc::METRIC_REC_TYPES);
-  }
+  HostGetStatusRequest request{{std::begin(data), std::end(data)}};
+
   auto response = invoke_rpc(request);
 
   _printer->write_output(response);

src/traffic_ctl_jsonrpc/CtrlPrinters.cc

@@ -282,12 +282,17 @@ RecordDescribePrinter::write_output_pretty(shared::rpc::RecordLookUpResponse con
 void
 GetHostStatusPrinter::write_output(YAML::Node const &result)
 {
-  auto response = result.as<shared::rpc::RecordLookUpResponse>();
-  for (auto &&recordInfo : response.recordList) {
-    std::cout << recordInfo.name << " " << recordInfo.currentValue << '\n';
+  auto resp = result.as<HostStatusLookUpResponse>();
+
+  if (resp.statusList.size() > 0) {
+    for (auto &&host : resp.statusList) {
+      std::cout << host.hostName << " " << host.status << '\n';
+    }
+    std::cout << '\n';
   }
-  for (auto &&e : response.errorList) {
-    std::cout << "Failed  to fetch " << e.recordName << '\n';
+
+  for (auto &&e : resp.errorList) {
+    std::cout << e << '\n';
   }
 }
 

src/traffic_ctl_jsonrpc/jsonrpc/CtrlRPCRequests.h

@@ -108,6 +108,16 @@ struct ConfigSetRecordResponse {
   std::vector<UpdatedRec> data;
 };
 //------------------------------------------------------------------------------------------------------------------------------------
+struct HostStatusLookUpResponse {
+  struct HostStatusInfo {
+    std::string hostName;
+    std::string status;
+  };
+
+  std::vector<HostStatusInfo> statusList;
+  std::vector<std::string> errorList;
+};
+//------------------------------------------------------------------------------------------------------------------------------------
 struct HostSetStatusRequest : shared::rpc::ClientRequest {
   using super = shared::rpc::ClientRequest;
   struct Params {
@@ -129,10 +139,16 @@ struct HostSetStatusRequest : shared::rpc::ClientRequest {
   }
 };
 
-struct HostGetStatusRequest : shared::rpc::RecordLookupRequest {
-  static constexpr auto STATUS_PREFIX = "proxy.process.host_status";
-  using super                         = shared::rpc::RecordLookupRequest;
-  HostGetStatusRequest() : super() {}
+struct HostGetStatusRequest : shared::rpc::ClientRequest {
+  using super  = shared::rpc::ClientRequest;
+  using Params = std::vector<std::string>;
+  HostGetStatusRequest(Params p) { super::params = std::move(p); }
+
+  std::string
+  get_method() const
+  {
+    return "admin_host_get_status";
+  }
 };
 //------------------------------------------------------------------------------------------------------------------------------------
 struct BasicPluginMessageRequest : shared::rpc::ClientRequest {

src/traffic_ctl_jsonrpc/jsonrpc/ctrl_yaml_codecs.h

@@ -157,5 +157,23 @@ template <> struct convert<ConfigSetRecordResponse> {
     return true;
   }
 };
-//------------------------------------------------------------------------------------------------------------------------------------
-} // namespace YAML
+
+template <> struct convert<HostStatusLookUpResponse> {
+  static bool
+  decode(Node const &node, HostStatusLookUpResponse &info)
+  {
+    YAML::Node statusList = node["statusList"];
+    YAML::Node errorList  = node["errorList"];
+    for (auto &&item : statusList) {
+      HostStatusLookUpResponse::HostStatusInfo hi;
+      hi.hostName = item["hostname"].Scalar();
+      hi.status   = item["status"].Scalar();
+      info.statusList.push_back(hi);
+    }
+    for (auto &&item : errorList) {
+      info.errorList.push_back(item.Scalar());
+    }
+    return true;
+  }
+};
+} // namespace YAML

src/traffic_ctl_jsonrpc/traffic_ctl.cc

@@ -93,7 +93,7 @@ main(int argc, const char **argv)
   config_command.add_command("registry", "Show configuration file registry", [&]() { command->execute(); })
     .add_example_usage("traffic_ctl config registry");
   // host commands
-  host_command.add_command("status", "Get one or more host statuses", "", MORE_THAN_ONE_ARG_N, [&]() { command->execute(); })
+  host_command.add_command("status", "Get one or more host statuses", "", MORE_THAN_ZERO_ARG_N, [&]() { command->execute(); })
     .add_example_usage("traffic_ctl host status HOST  [HOST  ...]");
   host_command.add_command("down", "Set down one or more host(s)", "", MORE_THAN_ONE_ARG_N, [&]() { command->execute(); })
     .add_example_usage("traffic_ctl host down HOST [OPTIONS]")