#include "common.h"
-int rrdpush_enabled = 0;
-int rrdpush_exclusive = 1;
+/*
+ * rrdpush
+ *
+ * 3 threads are involved for all stream operations
+ *
+ * 1. a random data collection thread, calling rrdset_done_push()
+ * this is called for each chart.
+ *
+ * the output of this work is kept in a BUFFER in RRDHOST
+ * the sender thread is signalled via a pipe (also in RRDHOST)
+ *
+ * 2. a sender thread running at the sending netdata
+ * this is spawned automatically on the first chart to be pushed
+ *
+ * It tries to push the metrics to the remote netdata, as fast
+ * as possible (i.e. immediately after they are collected).
+ *
+ * 3. a receiver thread, running at the receiving netdata
+ * this is spawned automatically when the sender connects to
+ * the receiver.
+ *
+ */
+
+#define START_STREAMING_PROMPT "Hit me baby, push them over..."
+
+int default_rrdpush_enabled = 0;
+char *default_rrdpush_destination = NULL;
+char *default_rrdpush_api_key = NULL;
-static char *remote_netdata_config = NULL;
-static char *api_key = NULL;
+int rrdpush_init() {
+ default_rrdpush_enabled = appconfig_get_boolean(&stream_config, CONFIG_SECTION_STREAM, "enabled", default_rrdpush_enabled);
+ default_rrdpush_destination = appconfig_get(&stream_config, CONFIG_SECTION_STREAM, "destination", "");
+ default_rrdpush_api_key = appconfig_get(&stream_config, CONFIG_SECTION_STREAM, "api key", "");
+ rrdhost_free_orphan_time = config_get_number(CONFIG_SECTION_GLOBAL, "cleanup orphan hosts after seconds", rrdhost_free_orphan_time);
+
+ if(default_rrdpush_enabled && (!default_rrdpush_destination || !*default_rrdpush_destination || !default_rrdpush_api_key || !*default_rrdpush_api_key)) {
+ error("STREAM [send]: cannot enable sending thread - information is missing.");
+ default_rrdpush_enabled = 0;
+ }
+
+ return default_rrdpush_enabled;
+}
#define CONNECTED_TO_SIZE 100
// that there are more data ready to be sent
#define PIPE_READ 0
#define PIPE_WRITE 1
-int rrdpush_pipe[2] = { -1, -1 };
-
-// a buffer used to store data to be sent.
-// the format is the same as external plugins.
-static BUFFER *rrdpush_buffer = NULL;
-
-// locking to get exclusive access to shared resources
-// (rrdpush_pipe[PIPE_WRITE], rrdpush_buffer
-static pthread_mutex_t rrdpush_mutex = PTHREAD_MUTEX_INITIALIZER;
-
-// if the streaming thread is connected to a remote netdata
-// this is set to 1, otherwise 0.
-static volatile int rrdpush_connected = 0;
// to have the remote netdata re-sync the charts
// to its current clock, we send for this many
// this is for the first iterations of each chart
static unsigned int remote_clock_resync_iterations = 60;
-#define rrdpush_lock() pthread_mutex_lock(&rrdpush_mutex)
-#define rrdpush_unlock() pthread_mutex_unlock(&rrdpush_mutex)
+#define rrdpush_lock(host) pthread_mutex_lock(&((host)->rrdpush_mutex))
+#define rrdpush_unlock(host) pthread_mutex_unlock(&((host)->rrdpush_mutex))
// checks if the current chart definition has been sent
static inline int need_to_send_chart_definition(RRDSET *st) {
RRDDIM *rd;
rrddim_foreach_read(rd, st)
- if(!rrddim_flag_check(rd, RRDDIM_FLAG_EXPOSED))
+ if(!rd->exposed)
return 1;
return 0;
// sends the current chart definition
static inline void send_chart_definition(RRDSET *st) {
- buffer_sprintf(rrdpush_buffer, "CHART '%s' '%s' '%s' '%s' '%s' '%s' '%s' %ld %d\n"
+ buffer_sprintf(st->rrdhost->rrdpush_buffer, "CHART '%s' '%s' '%s' '%s' '%s' '%s' '%s' %ld %d\n"
, st->id
, st->name
, st->title
RRDDIM *rd;
rrddim_foreach_read(rd, st) {
- buffer_sprintf(rrdpush_buffer, "DIMENSION '%s' '%s' '%s' " COLLECTED_NUMBER_FORMAT " " COLLECTED_NUMBER_FORMAT " '%s %s'\n"
+ buffer_sprintf(st->rrdhost->rrdpush_buffer, "DIMENSION '%s' '%s' '%s' " COLLECTED_NUMBER_FORMAT " " COLLECTED_NUMBER_FORMAT " '%s %s'\n"
, rd->id
, rd->name
, rrd_algorithm_name(rd->algorithm)
, rrddim_flag_check(rd, RRDDIM_FLAG_HIDDEN)?"hidden":""
, rrddim_flag_check(rd, RRDDIM_FLAG_DONT_DETECT_RESETS_OR_OVERFLOWS)?"noreset":""
);
- rrddim_flag_set(rd, RRDDIM_FLAG_EXPOSED);
+ rd->exposed = 1;
}
}
// sends the current chart dimensions
static inline void send_chart_metrics(RRDSET *st) {
- buffer_sprintf(rrdpush_buffer, "BEGIN %s %llu\n", st->id, (st->counter_done > remote_clock_resync_iterations)?st->usec_since_last_update:0);
+ buffer_sprintf(st->rrdhost->rrdpush_buffer, "BEGIN %s %llu\n", st->id, (st->counter_done > remote_clock_resync_iterations)?st->usec_since_last_update:0);
RRDDIM *rd;
rrddim_foreach_read(rd, st) {
- if(rrddim_flag_check(rd, RRDDIM_FLAG_UPDATED) && rrddim_flag_check(rd, RRDDIM_FLAG_EXPOSED))
- buffer_sprintf(rrdpush_buffer, "SET %s = " COLLECTED_NUMBER_FORMAT "\n"
+ if(rd->updated && rd->exposed)
+ buffer_sprintf(st->rrdhost->rrdpush_buffer, "SET %s = " COLLECTED_NUMBER_FORMAT "\n"
, rd->id
, rd->collected_value
);
}
- buffer_strcat(rrdpush_buffer, "END\n");
+ buffer_strcat(st->rrdhost->rrdpush_buffer, "END\n");
}
-// resets all the chart, so that their definitions
-// will be resent to the central netdata
-static void reset_all_charts(void) {
- rrd_rdlock();
-
- RRDHOST *host;
- rrdhost_foreach_read(host) {
- rrdhost_rdlock(host);
-
- RRDSET *st;
- rrdset_foreach_read(st, host) {
-
- // make it re-align the current time
- // on the remote host
- st->counter_done = 0;
-
- rrdset_rdlock(st);
-
- RRDDIM *rd;
- rrddim_foreach_read(rd, st)
- rrddim_flag_clear(rd, RRDDIM_FLAG_EXPOSED);
-
- rrdset_unlock(st);
- }
- rrdhost_unlock(host);
- }
- rrd_unlock();
-}
+void rrdpush_sender_thread_spawn(RRDHOST *host);
void rrdset_done_push(RRDSET *st) {
- static int error_shown = 0;
+ RRDHOST *host = st->rrdhost;
if(unlikely(!rrdset_flag_check(st, RRDSET_FLAG_ENABLED)))
return;
- rrdpush_lock();
+ rrdpush_lock(host);
+
+ if(unlikely(host->rrdpush_enabled && !host->rrdpush_spawn))
+ rrdpush_sender_thread_spawn(host);
- if(unlikely(!rrdpush_buffer || !rrdpush_connected)) {
- if(unlikely(!error_shown))
- error("STREAM [send]: not ready - discarding collected metrics.");
+ if(unlikely(!host->rrdpush_buffer || !host->rrdpush_connected)) {
+ if(unlikely(!host->rrdpush_error_shown))
+ error("STREAM %s [send]: not ready - discarding collected metrics.", host->hostname);
- error_shown = 1;
+ host->rrdpush_error_shown = 1;
- rrdpush_unlock();
+ rrdpush_unlock(host);
return;
}
- else if(unlikely(error_shown)) {
- error("STREAM [send]: ready - sending metrics...");
- error_shown = 0;
+ else if(unlikely(host->rrdpush_error_shown)) {
+ info("STREAM %s [send]: ready - sending metrics...", host->hostname);
+ host->rrdpush_error_shown = 0;
}
- rrdset_rdlock(st);
if(need_to_send_chart_definition(st))
send_chart_definition(st);
send_chart_metrics(st);
- rrdset_unlock(st);
// signal the sender there are more data
- if(write(rrdpush_pipe[PIPE_WRITE], " ", 1) == -1)
- error("STREAM [send]: cannot write to internal pipe");
+ if(write(host->rrdpush_pipe[PIPE_WRITE], " ", 1) == -1)
+ error("STREAM %s [send]: cannot write to internal pipe", host->hostname);
+
+ rrdpush_unlock(host);
+}
+
+// ----------------------------------------------------------------------------
+// rrdpush sender thread
- rrdpush_unlock();
+// resets all the chart, so that their definitions
+// will be resent to the central netdata
+static void rrdpush_sender_thread_reset_all_charts(RRDHOST *host) {
+ rrdhost_rdlock(host);
+
+ RRDSET *st;
+ rrdset_foreach_read(st, host) {
+
+ // make it re-align the current time
+ // on the remote host
+ st->counter_done = 0;
+
+ rrdset_rdlock(st);
+
+ RRDDIM *rd;
+ rrddim_foreach_read(rd, st)
+ rd->exposed = 0;
+
+ rrdset_unlock(st);
+ }
+
+ rrdhost_unlock(host);
}
-static inline void rrdpush_flush(void) {
- rrdpush_lock();
- if(buffer_strlen(rrdpush_buffer))
- error("STREAM [send]: discarding %zu bytes of metrics already in the buffer.", buffer_strlen(rrdpush_buffer));
+static inline void rrdpush_sender_thread_data_flush(RRDHOST *host) {
+ rrdpush_lock(host);
+
+ if(buffer_strlen(host->rrdpush_buffer))
+ error("STREAM %s [send]: discarding %zu bytes of metrics already in the buffer.", host->hostname, buffer_strlen(host->rrdpush_buffer));
+
+ buffer_flush(host->rrdpush_buffer);
- buffer_flush(rrdpush_buffer);
- reset_all_charts();
- rrdpush_unlock();
+ rrdpush_sender_thread_reset_all_charts(host);
+
+ rrdpush_unlock(host);
}
-int rrdpush_init() {
- rrdpush_enabled = config_get_boolean("stream", "enabled", rrdpush_enabled);
- rrdpush_exclusive = config_get_boolean("stream", "exclusive", rrdpush_exclusive);
- remote_netdata_config = config_get("stream", "stream metrics to", "");
- api_key = config_get("stream", "api key", "");
-
- if(!rrdpush_enabled || !remote_netdata_config || !*remote_netdata_config || !api_key || !*api_key) {
- rrdpush_enabled = 0;
- rrdpush_exclusive = 0;
+static void rrdpush_sender_thread_cleanup_locked_all(RRDHOST *host) {
+ host->rrdpush_connected = 0;
+
+ if(host->rrdpush_socket != -1) {
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
+ }
+
+ // close the pipe
+ if(host->rrdpush_pipe[PIPE_READ] != -1) {
+ close(host->rrdpush_pipe[PIPE_READ]);
+ host->rrdpush_pipe[PIPE_READ] = -1;
}
- return rrdpush_enabled;
+ if(host->rrdpush_pipe[PIPE_WRITE] != -1) {
+ close(host->rrdpush_pipe[PIPE_WRITE]);
+ host->rrdpush_pipe[PIPE_WRITE] = -1;
+ }
+
+ buffer_free(host->rrdpush_buffer);
+ host->rrdpush_buffer = NULL;
+
+ host->rrdpush_spawn = 0;
+
+ rrdhost_flag_set(host, RRDHOST_ORPHAN);
+}
+
+void rrdpush_sender_thread_stop(RRDHOST *host) {
+ rrdpush_lock(host);
+ rrdhost_wrlock(host);
+
+ if(host->rrdpush_spawn) {
+ info("STREAM %s [send]: stopping sending thread...", host->hostname);
+ pthread_cancel(host->rrdpush_thread);
+ rrdpush_sender_thread_cleanup_locked_all(host);
+ }
+
+ rrdhost_unlock(host);
+ rrdpush_unlock(host);
}
void *rrdpush_sender_thread(void *ptr) {
- struct netdata_static_thread *static_thread = (struct netdata_static_thread *)ptr;
+ RRDHOST *host = (RRDHOST *)ptr;
- info("STREAM [send]: thread created (task id %d)", gettid());
+ info("STREAM %s [send]: thread created (task id %d)", host->hostname, gettid());
if(pthread_setcanceltype(PTHREAD_CANCEL_DEFERRED, NULL) != 0)
- error("STREAM [send]: cannot set pthread cancel type to DEFERRED.");
+ error("STREAM %s [send]: cannot set pthread cancel type to DEFERRED.", host->hostname);
if(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL) != 0)
- error("STREAM [send]: cannot set pthread cancel state to ENABLE.");
-
- int timeout = (int)config_get_number("stream", "timeout seconds", 60);
- int default_port = (int)config_get_number("stream", "default port", 19999);
- size_t max_size = (size_t)config_get_number("stream", "buffer size bytes", 1024 * 1024);
- unsigned int reconnect_delay = (unsigned int)config_get_number("stream", "reconnect delay seconds", 5);
- remote_clock_resync_iterations = (unsigned int)config_get_number("stream", "initial clock resync iterations", remote_clock_resync_iterations);
- int sock = -1;
+ error("STREAM %s [send]: cannot set pthread cancel state to ENABLE.", host->hostname);
+
+ int timeout = (int)appconfig_get_number(&stream_config, CONFIG_SECTION_STREAM, "timeout seconds", 60);
+ int default_port = (int)appconfig_get_number(&stream_config, CONFIG_SECTION_STREAM, "default port", 19999);
+ size_t max_size = (size_t)appconfig_get_number(&stream_config, CONFIG_SECTION_STREAM, "buffer size bytes", 1024 * 1024);
+ unsigned int reconnect_delay = (unsigned int)appconfig_get_number(&stream_config, CONFIG_SECTION_STREAM, "reconnect delay seconds", 5);
+ remote_clock_resync_iterations = (unsigned int)appconfig_get_number(&stream_config, CONFIG_SECTION_STREAM, "initial clock resync iterations", remote_clock_resync_iterations);
char connected_to[CONNECTED_TO_SIZE + 1] = "";
- if(!rrdpush_enabled || !remote_netdata_config || !*remote_netdata_config || !api_key || !*api_key)
+ if(!host->rrdpush_enabled || !host->rrdpush_destination || !*host->rrdpush_destination || !host->rrdpush_api_key || !*host->rrdpush_api_key)
goto cleanup;
// initialize rrdpush globals
- rrdpush_buffer = buffer_create(1);
- rrdpush_connected = 0;
- if(pipe(rrdpush_pipe) == -1) fatal("STREAM [send]: cannot create required pipe.");
+ host->rrdpush_buffer = buffer_create(1);
+ host->rrdpush_connected = 0;
+ if(pipe(host->rrdpush_pipe) == -1) fatal("STREAM %s [send]: cannot create required pipe.", host->hostname);
// initialize local variables
size_t begin = 0;
ifd = &fds[0];
ofd = &fds[1];
- for(;;) {
- if(netdata_exit) break;
+ for(; host->rrdpush_enabled && !netdata_exit ;) {
- if(unlikely(sock == -1)) {
+ if(unlikely(host->rrdpush_socket == -1)) {
// stop appending data into rrdpush_buffer
// they will be lost, so there is no point to do it
- rrdpush_connected = 0;
+ host->rrdpush_connected = 0;
- info("STREAM [send to %s]: connecting...", remote_netdata_config);
- sock = connect_to_one_of(remote_netdata_config, default_port, &tv, &reconnects_counter, connected_to, CONNECTED_TO_SIZE);
+ info("STREAM %s [send to %s]: connecting...", host->hostname, host->rrdpush_destination);
+ host->rrdpush_socket = connect_to_one_of(host->rrdpush_destination, default_port, &tv, &reconnects_counter, connected_to, CONNECTED_TO_SIZE);
- if(unlikely(sock == -1)) {
- error("STREAM [send to %s]: failed to connect", remote_netdata_config);
+ if(unlikely(host->rrdpush_socket == -1)) {
+ error("STREAM %s [send to %s]: failed to connect", host->hostname, host->rrdpush_destination);
sleep(reconnect_delay);
continue;
}
- info("STREAM [send to %s]: initializing communication...", connected_to);
+ info("STREAM %s [send to %s]: initializing communication...", host->hostname, connected_to);
char http[1000 + 1];
snprintfz(http, 1000,
"STREAM key=%s&hostname=%s&machine_guid=%s&os=%s&update_every=%d HTTP/1.1\r\n"
"User-Agent: netdata-push-service/%s\r\n"
"Accept: */*\r\n\r\n"
- , api_key
- , localhost->hostname
- , localhost->machine_guid
- , localhost->os
+ , host->rrdpush_api_key
+ , host->hostname
+ , host->machine_guid
+ , host->os
, default_rrd_update_every
, program_version
);
- if(send_timeout(sock, http, strlen(http), 0, timeout) == -1) {
- close(sock);
- sock = -1;
- error("STREAM [send to %s]: failed to send http header to netdata", connected_to);
+ if(send_timeout(host->rrdpush_socket, http, strlen(http), 0, timeout) == -1) {
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
+ error("STREAM %s [send to %s]: failed to send http header to netdata", host->hostname, connected_to);
sleep(reconnect_delay);
continue;
}
- info("STREAM [send to %s]: waiting response from remote netdata...", connected_to);
+ info("STREAM %s [send to %s]: waiting response from remote netdata...", host->hostname, connected_to);
- if(recv_timeout(sock, http, 1000, 0, timeout) == -1) {
- close(sock);
- sock = -1;
- error("STREAM [send to %s]: failed to initialize communication", connected_to);
+ if(recv_timeout(host->rrdpush_socket, http, 1000, 0, timeout) == -1) {
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
+ error("STREAM %s [send to %s]: failed to initialize communication", host->hostname, connected_to);
sleep(reconnect_delay);
continue;
}
- if(strncmp(http, "STREAM", 6)) {
- close(sock);
- sock = -1;
- error("STREAM [send to %s]: server is not replying properly.", connected_to);
+ if(strncmp(http, START_STREAMING_PROMPT, strlen(START_STREAMING_PROMPT))) {
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
+ error("STREAM %s [send to %s]: server is not replying properly.", host->hostname, connected_to);
sleep(reconnect_delay);
continue;
}
- info("STREAM [send to %s]: established communication - sending metrics...", connected_to);
+ info("STREAM %s [send to %s]: established communication - sending metrics...", host->hostname, connected_to);
- if(fcntl(sock, F_SETFL, O_NONBLOCK) < 0)
- error("STREAM [send to %s]: cannot set non-blocking mode for socket.", connected_to);
+ if(fcntl(host->rrdpush_socket, F_SETFL, O_NONBLOCK) < 0)
+ error("STREAM %s [send to %s]: cannot set non-blocking mode for socket.", host->hostname, connected_to);
- rrdpush_flush();
+ rrdpush_sender_thread_data_flush(host);
sent_connection = 0;
// allow appending data into rrdpush_buffer
- rrdpush_connected = 1;
+ host->rrdpush_connected = 1;
}
- ifd->fd = rrdpush_pipe[PIPE_READ];
+ ifd->fd = host->rrdpush_pipe[PIPE_READ];
ifd->events = POLLIN;
ifd->revents = 0;
- ofd->fd = sock;
+ ofd->fd = host->rrdpush_socket;
ofd->revents = 0;
- if(begin < buffer_strlen(rrdpush_buffer)) {
+ if(begin < buffer_strlen(host->rrdpush_buffer)) {
ofd->events = POLLOUT;
fdmax = 2;
}
if(errno == EAGAIN || errno == EINTR)
continue;
- error("STREAM [send to %s]: failed to poll().", connected_to);
- close(sock);
- sock = -1;
+ error("STREAM %s [send to %s]: failed to poll().", host->hostname, connected_to);
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
break;
}
else if(unlikely(!retval)) {
if(ifd->revents & POLLIN) {
char buffer[1000 + 1];
- if(read(rrdpush_pipe[PIPE_READ], buffer, 1000) == -1)
- error("STREAM [send to %s]: cannot read from internal pipe.", connected_to);
+ if(read(host->rrdpush_pipe[PIPE_READ], buffer, 1000) == -1)
+ error("STREAM %s [send to %s]: cannot read from internal pipe.", host->hostname, connected_to);
}
- if(ofd->revents & POLLOUT && begin < buffer_strlen(rrdpush_buffer)) {
- rrdpush_lock();
- ssize_t ret = send(sock, &rrdpush_buffer->buffer[begin], buffer_strlen(rrdpush_buffer) - begin, MSG_DONTWAIT);
+ if(ofd->revents & POLLOUT && begin < buffer_strlen(host->rrdpush_buffer)) {
+
+ // BEGIN RRDPUSH LOCKED SESSION
+
+ // during this session, data collectors
+ // will not be able to append data to our buffer
+ // but the socket is in non-blocking mode
+ // so, we will not block at send()
+
+ if(pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, NULL) != 0)
+ error("STREAM %s [send]: cannot set pthread cancel state to DISABLE.", host->hostname);
+
+ rrdpush_lock(host);
+
+ ssize_t ret = send(host->rrdpush_socket, &host->rrdpush_buffer->buffer[begin], buffer_strlen(host->rrdpush_buffer) - begin, MSG_DONTWAIT);
if(ret == -1) {
if(errno != EAGAIN && errno != EINTR) {
- error("STREAM [send to %s]: failed to send metrics - closing connection - we have sent %zu bytes on this connection.", connected_to, sent_connection);
- close(sock);
- sock = -1;
+ error("STREAM %s [send to %s]: failed to send metrics - closing connection - we have sent %zu bytes on this connection.", host->hostname, connected_to, sent_connection);
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
}
}
else {
sent_connection += ret;
sent_bytes += ret;
begin += ret;
- if(begin == buffer_strlen(rrdpush_buffer)) {
- buffer_flush(rrdpush_buffer);
+ if(begin == buffer_strlen(host->rrdpush_buffer)) {
+ // we send it all
+
+ buffer_flush(host->rrdpush_buffer);
begin = 0;
}
}
- rrdpush_unlock();
+
+ rrdpush_unlock(host);
+
+ if(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL) != 0)
+ error("STREAM %s [send]: cannot set pthread cancel state to ENABLE.", host->hostname);
+
+ // END RRDPUSH LOCKED SESSION
}
// protection from overflow
- if(rrdpush_buffer->len > max_size) {
+ if(host->rrdpush_buffer->len > max_size) {
errno = 0;
- error("STREAM [send to %s]: too many data pending - buffer is %zu bytes long, %zu unsent - we have sent %zu bytes in total, %zu on this connection. Closing connection to flush the data.", connected_to, rrdpush_buffer->len, rrdpush_buffer->len - begin, sent_bytes, sent_connection);
- if(sock != -1) {
- close(sock);
- sock = -1;
+ error("STREAM %s [send to %s]: too many data pending - buffer is %zu bytes long, %zu unsent - we have sent %zu bytes in total, %zu on this connection. Closing connection to flush the data.", host->hostname, connected_to, host->rrdpush_buffer->len, host->rrdpush_buffer->len - begin, sent_bytes, sent_connection);
+ if(host->rrdpush_socket != -1) {
+ close(host->rrdpush_socket);
+ host->rrdpush_socket = -1;
}
}
}
cleanup:
- debug(D_WEB_CLIENT, "STREAM [send]: sending thread exits.");
-
- // make sure the data collection threads do not write data
- rrdpush_connected = 0;
+ debug(D_WEB_CLIENT, "STREAM %s [send]: sending thread exits.", host->hostname);
- // close the pipe
- if(rrdpush_pipe[PIPE_READ] != -1) close(rrdpush_pipe[PIPE_READ]);
- if(rrdpush_pipe[PIPE_WRITE] != -1) close(rrdpush_pipe[PIPE_WRITE]);
+ if(pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, NULL) != 0)
+ error("STREAM %s [send]: cannot set pthread cancel state to DISABLE.", host->hostname);
- // close the socket
- if(sock != -1) close(sock);
+ rrdpush_lock(host);
+ rrdhost_wrlock(host);
+ rrdpush_sender_thread_cleanup_locked_all(host);
+ rrdhost_unlock(host);
+ rrdpush_unlock(host);
- rrdpush_lock();
- buffer_free(rrdpush_buffer);
- rrdpush_buffer = NULL;
- rrdpush_unlock();
+ if(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL) != 0)
+ error("STREAM %s [send]: cannot set pthread cancel state to ENABLE.", host->hostname);
- static_thread->enabled = 0;
pthread_exit(NULL);
return NULL;
}
// ----------------------------------------------------------------------------
-// STREAM receiver
+// rrdpush receiver thread
int rrdpush_receive(int fd, const char *key, const char *hostname, const char *machine_guid, const char *os, int update_every, char *client_ip, char *client_port) {
RRDHOST *host;
int history = default_rrd_history_entries;
RRD_MEMORY_MODE mode = default_rrd_memory_mode;
int health_enabled = default_health_enabled;
+ int rrdpush_enabled = default_rrdpush_enabled;
+ char *rrdpush_destination = default_rrdpush_destination;
+ char *rrdpush_api_key = default_rrdpush_api_key;
time_t alarms_delay = 60;
update_every = (int)appconfig_get_number(&stream_config, machine_guid, "update every", update_every);
alarms_delay = appconfig_get_number(&stream_config, key, "default postpone alarms on connect seconds", alarms_delay);
alarms_delay = appconfig_get_number(&stream_config, machine_guid, "postpone alarms on connect seconds", alarms_delay);
+ rrdpush_enabled = appconfig_get_boolean(&stream_config, key, "default proxy enabled", rrdpush_enabled);
+ rrdpush_enabled = appconfig_get_boolean(&stream_config, machine_guid, "proxy enabled", rrdpush_enabled);
+
+ rrdpush_destination = appconfig_get(&stream_config, key, "default proxy destination", rrdpush_destination);
+ rrdpush_destination = appconfig_get(&stream_config, machine_guid, "proxy destination", rrdpush_destination);
+
+ rrdpush_api_key = appconfig_get(&stream_config, key, "default proxy api key", rrdpush_api_key);
+ rrdpush_api_key = appconfig_get(&stream_config, machine_guid, "proxy api key", rrdpush_api_key);
+
if(!strcmp(machine_guid, "localhost"))
host = localhost;
else
- host = rrdhost_find_or_create(hostname, machine_guid, os, update_every, history, mode, health_enabled?1:0);
-
- info("STREAM [receive from [%s]:%s]: metrics for host '%s' with machine_guid '%s': update every = %d, history = %d, memory mode = %s, health %s",
- client_ip, client_port,
- hostname, machine_guid,
- update_every,
- history,
- rrd_memory_mode_name(mode),
- (health_enabled == CONFIG_BOOLEAN_NO)?"disabled":((health_enabled == CONFIG_BOOLEAN_YES)?"enabled":"auto")
+ host = rrdhost_find_or_create(
+ hostname
+ , machine_guid
+ , os
+ , update_every
+ , history
+ , mode
+ , (health_enabled != CONFIG_BOOLEAN_NO)
+ , (rrdpush_enabled && rrdpush_destination && *rrdpush_destination && rrdpush_api_key && *rrdpush_api_key)
+ , rrdpush_destination
+ , rrdpush_api_key
+ );
+
+ if(!host) {
+ close(fd);
+ error("STREAM %s [receive from [%s]:%s]: failed to find/create host structure.", hostname, client_ip, client_port);
+ return 1;
+ }
+
+#ifdef NETDATA_INTERNAL_CHECKS
+ info("STREAM %s [receive from [%s]:%s]: client willing to stream metrics for host '%s' with machine_guid '%s': update every = %d, history = %d, memory mode = %s, health %s"
+ , hostname
+ , client_ip
+ , client_port
+ , host->hostname
+ , host->machine_guid
+ , host->rrd_update_every
+ , host->rrd_history_entries
+ , rrd_memory_mode_name(host->rrd_memory_mode)
+ , (health_enabled == CONFIG_BOOLEAN_NO)?"disabled":((health_enabled == CONFIG_BOOLEAN_YES)?"enabled":"auto")
);
+#endif // NETDATA_INTERNAL_CHECKS
struct plugind cd = {
.enabled = 1,
snprintfz(cd.fullfilename, FILENAME_MAX, "%s:%s", client_ip, client_port);
snprintfz(cd.cmd, PLUGINSD_CMD_MAX, "%s:%s", client_ip, client_port);
- info("STREAM [receive from [%s]:%s]: initializing communication...", client_ip, client_port);
- if(send_timeout(fd, "STREAM", 6, 0, 60) != 6) {
- error("STREAM [receive from [%s]:%s]: cannot send STREAM command.", client_ip, client_port);
+ info("STREAM %s [receive from [%s]:%s]: initializing communication...", host->hostname, client_ip, client_port);
+ if(send_timeout(fd, START_STREAMING_PROMPT, strlen(START_STREAMING_PROMPT), 0, 60) != strlen(START_STREAMING_PROMPT)) {
+ error("STREAM %s [receive from [%s]:%s]: cannot send ready command.", host->hostname, client_ip, client_port);
+ close(fd);
return 0;
}
// remove the non-blocking flag from the socket
if(fcntl(fd, F_SETFL, fcntl(fd, F_GETFL, 0) & ~O_NONBLOCK) == -1)
- error("STREAM [receive from [%s]:%s]: cannot remove the non-blocking flag from socket %d", client_ip, client_port, fd);
+ error("STREAM %s [receive from [%s]:%s]: cannot remove the non-blocking flag from socket %d", host->hostname, client_ip, client_port, fd);
// convert the socket to a FILE *
FILE *fp = fdopen(fd, "r");
if(!fp) {
- error("STREAM [receive from [%s]:%s]: failed to get a FILE for FD %d.", client_ip, client_port, fd);
+ error("STREAM %s [receive from [%s]:%s]: failed to get a FILE for FD %d.", host->hostname, client_ip, client_port, fd);
+ close(fd);
return 0;
}
rrdhost_wrlock(host);
- host->use_counter++;
+ host->connected_senders++;
if(health_enabled != CONFIG_BOOLEAN_NO)
host->health_delay_up_to = now_realtime_sec() + alarms_delay;
rrdhost_unlock(host);
// call the plugins.d processor to receive the metrics
- info("STREAM [receive from [%s]:%s]: connecting client to plugins.d (host '%s', machine GUID '%s').", client_ip, client_port, host->hostname, host->machine_guid);
+ info("STREAM %s [receive from [%s]:%s]: receiving metrics...", host->hostname, client_ip, client_port);
size_t count = pluginsd_process(host, &cd, fp, 1);
- error("STREAM [receive from [%s]:%s]: client disconnected (host '%s', machine GUID '%s', completed updates %zu).", client_ip, client_port, host->hostname, host->machine_guid, count);
+ error("STREAM %s [receive from [%s]:%s]: disconnected (completed updates %zu).", host->hostname, client_ip, client_port, count);
rrdhost_wrlock(host);
- host->use_counter--;
- if(!host->use_counter && health_enabled == CONFIG_BOOLEAN_AUTO)
- host->health_enabled = 0;
+ host->connected_senders--;
+ if(!host->connected_senders) {
+ if(health_enabled == CONFIG_BOOLEAN_AUTO)
+ host->health_enabled = 0;
+
+ host->senders_disconnected_time = now_realtime_sec();
+ }
rrdhost_unlock(host);
+ rrdpush_sender_thread_stop(host);
+
// cleanup
fclose(fp);
struct rrdpush_thread *rpt = (struct rrdpush_thread *)ptr;
if (pthread_setcanceltype(PTHREAD_CANCEL_DEFERRED, NULL) != 0)
- error("STREAM [receive]: cannot set pthread cancel type to DEFERRED.");
+ error("STREAM %s [receive]: cannot set pthread cancel type to DEFERRED.", rpt->hostname);
if (pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL) != 0)
- error("STREAM [receive]: cannot set pthread cancel state to ENABLE.");
+ error("STREAM %s [receive]: cannot set pthread cancel state to ENABLE.", rpt->hostname);
- info("STREAM [%s]:%s: receive thread created (task id %d)", rpt->client_ip, rpt->client_port, gettid());
+ info("STREAM %s [%s]:%s: receive thread created (task id %d)", rpt->hostname, rpt->client_ip, rpt->client_port, gettid());
rrdpush_receive(rpt->fd, rpt->key, rpt->hostname, rpt->machine_guid, rpt->os, rpt->update_every, rpt->client_ip, rpt->client_port);
- info("STREAM [receive from [%s]:%s]: receive thread ended (task id %d)", rpt->client_ip, rpt->client_port, gettid());
+ info("STREAM %s [receive from [%s]:%s]: receive thread ended (task id %d)", rpt->hostname, rpt->client_ip, rpt->client_port, gettid());
- close(rpt->fd);
freez(rpt->key);
freez(rpt->hostname);
freez(rpt->machine_guid);
return NULL;
}
-static inline int rrdpush_receive_validate_api_key(const char *key) {
- return appconfig_get_boolean(&stream_config, key, "enabled", 0);
+void rrdpush_sender_thread_spawn(RRDHOST *host) {
+ rrdhost_wrlock(host);
+
+ if(!host->rrdpush_spawn) {
+ if(pthread_create(&host->rrdpush_thread, NULL, rrdpush_sender_thread, (void *) host))
+ error("STREAM %s [send]: failed to create new thread for client.", host->hostname);
+
+ else if(pthread_detach(host->rrdpush_thread))
+ error("STREAM %s [send]: cannot request detach newly created thread.", host->hostname);
+
+ rrdhost_flag_clear(host, RRDHOST_ORPHAN);
+ host->rrdpush_spawn = 1;
+ }
+
+ rrdhost_unlock(host);
}
int rrdpush_receiver_thread_spawn(RRDHOST *host, struct web_client *w, char *url) {
(void)host;
-
+
info("STREAM [receive from [%s]:%s]: new client connection.", w->client_ip, w->client_port);
- char *key = NULL, *hostname = NULL, *machine_guid = NULL, *os = NULL;
+ char *key = NULL, *hostname = NULL, *machine_guid = NULL, *os = "unknown";
int update_every = default_rrd_update_every;
+ char buf[GUID_LEN + 1];
while(url) {
char *value = mystrsep(&url, "?&");
return 400;
}
- if(!rrdpush_receive_validate_api_key(key)) {
- error("STREAM [receive from [%s]:%s]: API key '%s' is not allowed. Forbidding access.", w->client_ip, w->client_port, key);
+ if(regenerate_guid(key, buf) == -1) {
+ error("STREAM [receive from [%s]:%s]: API key '%s' is not valid GUID. Forbidding access.", w->client_ip, w->client_port, key);
+ buffer_flush(w->response.data);
+ buffer_sprintf(w->response.data, "Your API key is invalid.");
+ return 401;
+ }
+
+ if(regenerate_guid(machine_guid, buf) == -1) {
+ error("STREAM [receive from [%s]:%s]: machine GUID '%s' is not GUID. Forbidding access.", w->client_ip, w->client_port, key);
+ buffer_flush(w->response.data);
+ buffer_sprintf(w->response.data, "Your machine GUID is invalid.");
+ return 404;
+ }
+
+ if(!appconfig_get_boolean(&stream_config, key, "enabled", 0)) {
+ error("STREAM [receive from [%s]:%s]: API key '%s' is not allowed. Forbidding access.", w->client_ip, w->client_port, machine_guid);
buffer_flush(w->response.data);
buffer_sprintf(w->response.data, "Your API key is not permitted access.");
return 401;