CPP-964 Add refresh-interval support for histogram metrics

[absurdfarce]

Port of Java functionality as defined here.

[absurdfarce]

Some sample code for playing with the refresh rate in a real app:

#include "cassandra.h"
/* Use "#include <dse.h>" when connecting to DataStax Enterpise */
#include <stdio.h>
#include <unistd.h>

void on_log(const CassLogMessage* message, void* data) {
  FILE* log_file = (FILE*)data;
  fprintf(log_file, "%u.%03u [%s] (%s:%d:%s): %s\n", (unsigned int)(message->time_ms / 1000),
          (unsigned int)(message->time_ms % 1000), cass_log_level_string(message->severity),
          message->file, message->line, message->function, message->message);
}

void execute_query(CassStatement* statement, CassSession* session) {

    CassFuture* result_future = cass_session_execute(session, statement);

    if (cass_future_error_code(result_future) == CASS_OK) {
      /* Retrieve result set and get the first row */
      const CassResult* result = cass_future_get_result(result_future);
      const CassRow* row = cass_result_first_row(result);

      if (row) {
        const CassValue* value = cass_row_get_column_by_name(row, "release_version");

        const char* release_version;
        size_t release_version_length;
        cass_value_get_string(value, &release_version, &release_version_length);
        printf("release_version: '%.*s'\n", (int)release_version_length, release_version);
      }

      cass_result_free(result);
    } else {
      /* Handle error */
      const char* message;
      size_t message_length;
      cass_future_error_message(result_future, &message, &message_length);
      fprintf(stderr, "Unable to run query: '%.*s'\n", (int)message_length, message);
    }

    cass_future_free(result_future);
}

int main(int argc, char* argv[]) {

  FILE* log_file = fopen("driver.log", "w+");
  if (log_file == NULL) {
    fprintf(stderr, "Unable to open log file\n");
  }

  /* Log configuration *MUST* be done before any other driver call */
  cass_log_set_level(CASS_LOG_INFO);
  cass_log_set_callback(on_log, (void*)log_file);

  /* Setup and connect to cluster */
  CassFuture* connect_future = NULL;
  CassCluster* cluster = cass_cluster_new();
  CassSession* session = cass_session_new();
  const char* hosts = "127.0.0.1";
  if (argc > 1) {
    hosts = argv[1];
  }
  
  /* Add contact points */
  cass_cluster_set_contact_points(cluster, hosts);

  /* Specify a histogram refresh period */
  cass_cluster_set_histogram_refresh_interval(cluster, 5000);

  /* Provide the cluster object as configuration to connect the session */
  connect_future = cass_session_connect(session, cluster);

  if (cass_future_error_code(connect_future) == CASS_OK) {
    CassFuture* close_future = NULL;

    /* Build statement and execute query */
    const char* query = "SELECT release_version FROM system.local";
    CassStatement* statement = cass_statement_new(query, 0);
    CassMetrics metrics;

    for (int i = 1; i < 10; ++i) {

      execute_query(statement, session);
      sleep(1);

      cass_session_get_metrics(session, &metrics);
      printf("Requests min: %u\n", metrics.requests.min);
    }

    /* Close the session */
    cass_statement_free(statement);
    close_future = cass_session_close(session);
    cass_future_wait(close_future);
    cass_future_free(close_future);
  } else {
    /* Handle error */
    const char* message;
    size_t message_length;
    cass_future_error_message(connect_future, &message, &message_length);
    fprintf(stderr, "Unable to connect: '%.*s'\n", (int)message_length, message);
  }

  cass_future_free(connect_future);
  cass_cluster_free(cluster);
  cass_session_free(session);

  return 0;
}

[absurdfarce]

I don't love this name for the constant but... I couldn't come up with something I liked better.

[absurdfarce]

This feels like something that should be a static value on the Histogram class but repeated attempts to setup such a value hit various issues that my knowledge of C++ wasn't subtle enough to address. Since we're only creating one of these for a given metrics impl I wasn't too worried about it (the size of a Snapshot shouldn't be a driving concern) but I concede it's not ideal.

[absurdfarce]

Model here very much is the impl of similar functionality in the 4.x Java driver

[hhughes]

Nit: can we use build_and_copy_snapshot(histogram_, &cached_snapshot_) here to save an object creation?

[absurdfarce]

Not obviously. The whole point of build_and_copy_snapshot() was to avoid the creation of an intermediate Snapshot object and just update the "rv" Snapshot* pointer directly. In this line we very explicitly want to create that intermediate Snapshot object because then we're going to cache it.

[joao-r-reis]

Left a NIT comment, I don't have time to set up an env to build this locally and I see that Henry is already doing that so I'll leave that part to him for now 😅

[absurdfarce]

Hey @hhughes and/or @joao-r-reis ; you guys good after the last round of changes? I think everything has been addressed but wanted a final confirmation from one or (ideally) both of you on that point.

[joao-r-reis]

Yeah the issue I brought up is resolved.

[hhughes]

LGTM