#define BACKEND_SOURCE_DATA_AS_COLLECTED 0x00000001
#define BACKEND_SOURCE_DATA_AVERAGE 0x00000002
-#define BACKEND_SOURCE_DATA_SUM 0x00000003
+#define BACKEND_SOURCE_DATA_SUM 0x00000004
-int connect_to_socket4(const char *ip, int port) {
+int connect_to_socket4(const char *ip, int port, struct timeval *timeout) {
int sock;
debug(D_LISTENER, "IPv4 connecting to ip '%s' port %d", ip, port);
return -1;
}
+ if(setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (char *)timeout, sizeof(struct timeval)) < 0)
+ error("Failed to set timeout on the socket to ip '%s' port %d", ip, port);
+
struct sockaddr_in name;
memset(&name, 0, sizeof(struct sockaddr_in));
name.sin_family = AF_INET;
return sock;
}
-int connect_to_socket6(const char *ip, int port) {
+int connect_to_socket6(const char *ip, int port, struct timeval *timeout) {
int sock = -1;
int ipv6only = 1;
return -1;
}
+ if(setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (char *)timeout, sizeof(struct timeval)) < 0)
+ error("Failed to set timeout on the socket to ip '%s' port %d", ip, port);
+
/* IPv6 only */
if(setsockopt(sock, IPPROTO_IPV6, IPV6_V6ONLY, (void*)&ipv6only, sizeof(ipv6only)) != 0)
error("Cannot set IPV6_V6ONLY on ip '%s' port's %d.", ip, port);
}
-static inline int connect_to_one(const char *definition, int default_port) {
+static inline int connect_to_one(const char *definition, int default_port, struct timeval *timeout) {
struct addrinfo hints;
struct addrinfo *result = NULL, *rp = NULL;
struct sockaddr_in *sin = (struct sockaddr_in *) rp->ai_addr;
inet_ntop(AF_INET, &sin->sin_addr, rip, INET_ADDRSTRLEN);
rport = ntohs(sin->sin_port);
- fd = connect_to_socket4(rip, rport);
+ fd = connect_to_socket4(rip, rport, timeout);
break;
}
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *) rp->ai_addr;
inet_ntop(AF_INET6, &sin6->sin6_addr, rip, INET6_ADDRSTRLEN);
rport = ntohs(sin6->sin6_port);
- fd = connect_to_socket6(rip, rport);
+ fd = connect_to_socket6(rip, rport, timeout);
break;
}
}
return fd;
}
-static inline calculated_number backend_duration_average(RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
+static inline calculated_number backend_calculate_value_from_stored_data(RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
time_t first_t = rrdset_first_entry_t(st);
time_t last_t = rrdset_last_entry_t(st);
return sum / (calculated_number)counter;
}
-static inline void format_dimension_collected_graphite_plaintext(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
+static inline int format_dimension_collected_graphite_plaintext(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
(void)host;
(void)after;
(void)before;
(void)options;
buffer_sprintf(b, "%s.%s.%s.%s " COLLECTED_NUMBER_FORMAT " %u\n", prefix, hostname, st->id, rd->id, rd->last_collected_value, (uint32_t)rd->last_collected_time.tv_sec);
+ return 1;
}
-static inline void format_dimension_stored_graphite_plaintext(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
+static inline int format_dimension_stored_graphite_plaintext(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
(void)host;
- calculated_number value = backend_duration_average(st, rd, after, before, options);
- if(!isnan(value))
- buffer_sprintf(b, "%s.%s.%s.%s " CALCULATED_NUMBER_FORMAT " %u\n", prefix, hostname, st->id, rd->id, value, (uint32_t)before);
+ calculated_number value = backend_calculate_value_from_stored_data(st, rd, after, before, options);
+ if(!isnan(value)) {
+ buffer_sprintf(b, "%s.%s.%s.%s " CALCULATED_NUMBER_FORMAT " %u\n", prefix, hostname, st->id, rd->id, value, (uint32_t) before);
+ return 1;
+ }
+ return 0;
}
-static inline void format_dimension_collected_opentsdb_telnet(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
+static inline int format_dimension_collected_opentsdb_telnet(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
(void)host;
(void)after;
(void)before;
(void)options;
buffer_sprintf(b, "put %s.%s.%s %u " COLLECTED_NUMBER_FORMAT " host=%s\n", prefix, st->id, rd->id, (uint32_t)rd->last_collected_time.tv_sec, rd->last_collected_value, hostname);
+ return 1;
}
-static inline void format_dimension_stored_opentsdb_telnet(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
+static inline int format_dimension_stored_opentsdb_telnet(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) {
(void)host;
- calculated_number value = backend_duration_average(st, rd, after, before, options);
- if(!isnan(value))
- buffer_sprintf(b, "put %s.%s.%s %u " CALCULATED_NUMBER_FORMAT " host=%s\n", prefix, st->id, rd->id, (uint32_t)before, value, hostname);
+ calculated_number value = backend_calculate_value_from_stored_data(st, rd, after, before, options);
+ if(!isnan(value)) {
+ buffer_sprintf(b, "put %s.%s.%s %u " CALCULATED_NUMBER_FORMAT " host=%s\n", prefix, st->id, rd->id, (uint32_t) before, value, hostname);
+ return 1;
+ }
+ return 0;
}
void *backends_main(void *ptr) {
(void)ptr;
BUFFER *b = buffer_create(1);
- void (*formatter)(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options);
+ int (*formatter)(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options);
info("BACKEND thread created with task id %d", gettid());
if(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL) != 0)
error("Cannot set pthread cancel state to ENABLE.");
+ // ------------------------------------------------------------------------
+ // collect configuration options
+
+ struct timeval timeout = {
+ .tv_sec = 0,
+ .tv_usec = 0
+ };
int default_port = 0;
int sock = -1;
- uint32_t options = BACKEND_SOURCE_DATA_AVERAGE;
- int enabled = config_get_boolean("backend", "enable", 0);
+ uint32_t options;
+ int enabled = config_get_boolean("backend", "enabled", 0);
const char *source = config_get("backend", "data source", "average");
const char *type = config_get("backend", "type", "graphite");
const char *destination = config_get("backend", "destination", "localhost");
const char *prefix = config_get("backend", "prefix", "netdata");
const char *hostname = config_get("backend", "hostname", localhost.hostname);
- int frequency = (int)config_get_number("backend", "update every", 30);
+ int frequency = (int)config_get_number("backend", "update every", 10);
int buffer_on_failures = (int)config_get_number("backend", "buffer on failures", 10);
+ long timeoutms = config_get_number("backend", "timeout ms", frequency * 2 * 1000);
+ // ------------------------------------------------------------------------
+ // validate configuration options
+ // and prepare for sending data to our backend
if(!enabled || frequency < 1)
goto cleanup;
goto cleanup;
}
+ if(timeoutms < 1) {
+ error("BACKED invalid timeout %ld ms given. Assuming %d ms.", timeoutms, frequency * 2 * 1000);
+ timeoutms = frequency * 2 * 1000;
+ }
+ timeout.tv_sec = (timeoutms * 1000) / 1000000;
+ timeout.tv_usec = (timeoutms * 1000) % 1000000;
+
+ // ------------------------------------------------------------------------
+ // prepare the charts for monitoring the backend
+
+ struct rusage thread;
+
+ collected_number
+ chart_buffered_metrics = 0,
+ chart_lost_metrics = 0,
+ chart_sent_metrics = 0,
+ chart_buffered_bytes = 0,
+ chart_sent_bytes = 0,
+ chart_transmission_successes = 0,
+ chart_transmission_failures = 0,
+ chart_data_lost_events = 0,
+ chart_lost_bytes = 0,
+ chart_backend_reconnects = 0,
+ chart_backend_latency = 0;
+
+ RRDSET *chart_metrics = rrdset_find("netdata.backend_metrics");
+ if(!chart_metrics) {
+ chart_metrics = rrdset_create("netdata", "backend_metrics", NULL, "backend", NULL, "Netdata Buffered Metrics", "metrics", 130600, frequency, RRDSET_TYPE_LINE);
+ rrddim_add(chart_metrics, "buffered", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_metrics, "lost", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_metrics, "sent", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ }
+
+ RRDSET *chart_bytes = rrdset_find("netdata.backend_bytes");
+ if(!chart_bytes) {
+ chart_bytes = rrdset_create("netdata", "backend_bytes", NULL, "backend", NULL, "Netdata Backend Data Size", "KB", 130610, frequency, RRDSET_TYPE_AREA);
+ rrddim_add(chart_bytes, "buffered", NULL, 1, 1024, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_bytes, "lost", NULL, 1, 1024, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_bytes, "sent", NULL, 1, 1024, RRDDIM_ABSOLUTE);
+ }
+
+ RRDSET *chart_ops = rrdset_find("netdata.backend_ops");
+ if(!chart_ops) {
+ chart_ops = rrdset_create("netdata", "backend_ops", NULL, "backend", NULL, "Netdata Backend Operations", "operations", 130630, frequency, RRDSET_TYPE_LINE);
+ rrddim_add(chart_ops, "write", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_ops, "discard", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_ops, "reconnect", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ rrddim_add(chart_ops, "failure", NULL, 1, 1, RRDDIM_ABSOLUTE);
+ }
+
+ RRDSET *chart_latency = rrdset_find("netdata.backend_latency");
+ if(!chart_latency) {
+ chart_latency = rrdset_create("netdata", "backend_latency", NULL, "backend", NULL, "Netdata Backend Latency", "ms", 130620, frequency, RRDSET_TYPE_AREA);
+ rrddim_add(chart_latency, "latency", NULL, 1, 1000, RRDDIM_ABSOLUTE);
+ }
+
+ RRDSET *chart_rusage = rrdset_find("netdata.backend_thread_cpu");
+ if(!chart_rusage) {
+ chart_rusage = rrdset_create("netdata", "backend_thread_cpu", NULL, "backend", NULL, "NetData Backend Thread CPU usage", "milliseconds/s", 130630, frequency, RRDSET_TYPE_STACKED);
+ rrddim_add(chart_rusage, "user", NULL, 1, 1000, RRDDIM_INCREMENTAL);
+ rrddim_add(chart_rusage, "system", NULL, 1, 1000, RRDDIM_INCREMENTAL);
+ }
+
+ // ------------------------------------------------------------------------
+ // prepare the backend main loop
+
+ info("BACKEND configured ('%s' on '%s' sending '%s' data, every %d seconds, as host '%s', with prefix '%s')", type, destination, source, frequency, hostname, prefix);
+
unsigned long long step_ut = frequency * 1000000ULL;
- unsigned long long random_ut = time_usec() % (step_ut / 2);
- time_t before = (time_t)((time_usec() - step_ut) / 10000000ULL);
+ unsigned long long random_ut = now_realtime_usec() % (step_ut / 2);
+ time_t before = (time_t)((now_realtime_usec() - step_ut) / 10000000ULL);
time_t after = before;
int failures = 0;
- info("BACKEND configured ('%s' on '%s' sending '%s' data, every %d seconds, as host '%s', with prefix '%s')", type, destination, source, frequency, hostname, prefix);
-
for(;;) {
- unsigned long long now_ut = time_usec();
+ // ------------------------------------------------------------------------
+ // wait for the next iteration point
+
+ unsigned long long now_ut = now_realtime_usec();
unsigned long long next_ut = now_ut - (now_ut % step_ut) + step_ut;
before = (time_t)(next_ut / 1000000ULL);
while(now_ut < next_ut) {
sleep_usec(next_ut - now_ut);
- now_ut = time_usec();
+ now_ut = now_realtime_usec();
}
+ // ------------------------------------------------------------------------
+ // add to the buffer the data we need to send to the backend
+
RRDSET *st;
int pthreadoldcancelstate;
RRDDIM *rd;
for(rd = st->dimensions; rd ;rd = rd->next) {
if(rd->last_collected_time.tv_sec >= after)
- formatter(b, prefix, &localhost, hostname, st, rd, after, before, options);
+ chart_buffered_metrics += formatter(b, prefix, &localhost, hostname, st, rd, after, before, options);
}
pthread_rwlock_unlock(&st->rwlock);
if(unlikely(pthread_setcancelstate(pthreadoldcancelstate, NULL) != 0))
error("Cannot set pthread cancel state to RESTORE (%d).", pthreadoldcancelstate);
+ chart_buffered_bytes = (collected_number)buffer_strlen(b);
+
+ // reset the monitoring chart counters
+ chart_sent_bytes =
+ chart_sent_metrics =
+ chart_lost_metrics =
+ chart_transmission_successes =
+ chart_transmission_failures =
+ chart_data_lost_events =
+ chart_lost_bytes =
+ chart_backend_reconnects =
+ chart_backend_latency = 0;
+
if(unlikely(netdata_exit)) break;
- //fprintf(stderr, "\nBACKEND BEGIN:\n%s\nBACKEND END\n", buffer_tostring(b));
+ //fprintf(stderr, "\nBACKEND BEGIN:\n%s\nBACKEND END\n", buffer_tostring(b)); // FIXME
//fprintf(stderr, "after = %lu, before = %lu\n", after, before);
+ // ------------------------------------------------------------------------
+ // connect to a backend server
+
if(unlikely(sock == -1)) {
+ unsigned long long start_ut = now_realtime_usec();
const char *s = destination;
while(*s) {
const char *e = s;
char buf[e - s + 1];
strncpyz(buf, s, e - s);
- sock = connect_to_one(buf, default_port);
+ chart_backend_reconnects++;
+ sock = connect_to_one(buf, default_port, &timeout);
if(sock != -1) break;
s = e;
}
+ chart_backend_latency += now_realtime_usec() - start_ut;
}
if(unlikely(netdata_exit)) break;
+ // ------------------------------------------------------------------------
+ // send our buffer to the backend server
+
if(likely(sock != -1)) {
size_t len = buffer_strlen(b);
- ssize_t written = write(sock, buffer_tostring(b), len);
+ unsigned long long start_ut = now_realtime_usec();
+ int flags = 0;
+#ifdef MSG_NOSIGNAL
+ flags += MSG_NOSIGNAL;
+#endif
+ ssize_t written = send(sock, buffer_tostring(b), len, flags);
+ chart_backend_latency += now_realtime_usec() - start_ut;
if(written != -1 && (size_t)written == len) {
+ // we sent the data successfully
+ chart_transmission_successes++;
+ chart_sent_bytes += written;
+ chart_sent_metrics = chart_buffered_metrics;
+
+ // reset the failures count
failures = 0;
+
+ // empty the buffer
buffer_flush(b);
}
else {
- failures++;
+ // oops! we couldn't send (all or some of the) data
error("Failed to write data to database backend '%s'. Willing to write %zu bytes, wrote %zd bytes. Will re-connect.", destination, len, written);
+ chart_transmission_failures++;
+
+ if(written != -1)
+ chart_sent_bytes += written;
+
+ // increment the counter we check for data loss
+ failures++;
+
+ // close the socket - we will re-open it next time
close(sock);
sock = -1;
}
+ // either the buffer is empty
+ // or is holding the data we couldn't send
+ // so, make sure the next iteration will continue
+ // from where we are now
after = before;
}
else {
- failures++;
error("Failed to update database backend '%s'", destination);
+ chart_transmission_failures++;
+
+ // increment the counter we check for data loss
+ failures++;
}
if(failures > buffer_on_failures) {
+ // too bad! we are going to lose data
+ chart_lost_bytes += buffer_strlen(b);
error("Reached %d backend failures. Flushing buffers to protect this host - this results in data loss on back-end server '%s'", failures, destination);
buffer_flush(b);
failures = 0;
+ chart_data_lost_events++;
+ chart_lost_metrics = chart_buffered_metrics;
}
if(unlikely(netdata_exit)) break;
+
+ // ------------------------------------------------------------------------
+ // update the monitoring charts
+
+ if(chart_ops->counter_done) rrdset_next(chart_ops);
+ rrddim_set(chart_ops, "write", chart_transmission_successes);
+ rrddim_set(chart_ops, "discard", chart_data_lost_events);
+ rrddim_set(chart_ops, "failure", chart_transmission_failures);
+ rrddim_set(chart_ops, "reconnect", chart_backend_reconnects);
+ rrdset_done(chart_ops);
+
+ if(chart_metrics->counter_done) rrdset_next(chart_metrics);
+ rrddim_set(chart_metrics, "buffered", chart_buffered_metrics);
+ rrddim_set(chart_metrics, "lost", chart_lost_metrics);
+ rrddim_set(chart_metrics, "sent", chart_sent_metrics);
+ rrdset_done(chart_metrics);
+
+ if(chart_bytes->counter_done) rrdset_next(chart_bytes);
+ rrddim_set(chart_bytes, "buffered", chart_buffered_bytes);
+ rrddim_set(chart_bytes, "lost", chart_lost_bytes);
+ rrddim_set(chart_bytes, "sent", chart_sent_bytes);
+ rrdset_done(chart_bytes);
+
+ if(chart_latency->counter_done) rrdset_next(chart_latency);
+ rrddim_set(chart_latency, "latency", chart_backend_latency);
+ rrdset_done(chart_latency);
+
+ getrusage(RUSAGE_THREAD, &thread);
+ if(chart_rusage->counter_done) rrdset_next(chart_rusage);
+ rrddim_set(chart_rusage, "user", thread.ru_utime.tv_sec * 1000000ULL + thread.ru_utime.tv_usec);
+ rrddim_set(chart_rusage, "system", thread.ru_stime.tv_sec * 1000000ULL + thread.ru_stime.tv_usec);
+ rrdset_done(chart_rusage);
+
+ if(likely(buffer_strlen(b) == 0))
+ chart_buffered_metrics = 0;
+
+ if(unlikely(netdata_exit)) break;
}
cleanup:
projects / netdata.git / blobdiff