#define BACKEND_SOURCE_DATA_AVERAGE 0x00000002
#define BACKEND_SOURCE_DATA_SUM 0x00000004
-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);
-
- sock = socket(AF_INET, SOCK_STREAM, 0);
- if(sock < 0) {
- error("IPv4 socket() on ip '%s' port %d failed.", 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;
- name.sin_port = htons(port);
-
- int ret = inet_pton(AF_INET, ip, (void *)&name.sin_addr.s_addr);
- if(ret != 1) {
- error("Failed to convert '%s' to a valid IPv4 address.", ip);
- close(sock);
- return -1;
- }
-
- if(connect(sock, (struct sockaddr *) &name, sizeof(name)) < 0) {
- close(sock);
- error("IPv4 failed to connect to '%s', port %d", ip, port);
- return -1;
- }
-
- debug(D_LISTENER, "Connected to IPv4 ip '%s' port %d", ip, port);
- return sock;
-}
-
-int connect_to_socket6(const char *ip, int port, struct timeval *timeout) {
- int sock = -1;
- int ipv6only = 1;
-
- debug(D_LISTENER, "IPv6 connecting to ip '%s' port %d", ip, port);
-
- sock = socket(AF_INET6, SOCK_STREAM, 0);
- if (sock < 0) {
- error("IPv6 socket() on ip '%s' port %d failed.", 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);
-
- /* 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);
-
- struct sockaddr_in6 name;
- memset(&name, 0, sizeof(struct sockaddr_in6));
- name.sin6_family = AF_INET6;
- name.sin6_port = htons ((uint16_t) port);
-
- int ret = inet_pton(AF_INET6, ip, (void *)&name.sin6_addr.s6_addr);
- if(ret != 1) {
- error("Failed to convert IP '%s' to a valid IPv6 address.", ip);
- close(sock);
- return -1;
- }
-
- name.sin6_scope_id = 0;
-
- if(connect(sock, (struct sockaddr *)&name, sizeof(name)) < 0) {
- close(sock);
- error("IPv6 failed to connect to '%s', port %d", ip, port);
- return -1;
- }
-
- debug(D_LISTENER, "Connected to IPv6 ip '%s' port %d", ip, port);
- return sock;
-}
-
-
-static inline int connect_to_one(const char *definition, int default_port, struct timeval *timeout) {
- struct addrinfo hints;
- struct addrinfo *result = NULL, *rp = NULL;
-
- char buffer[strlen(definition) + 1];
- strcpy(buffer, definition);
-
- char buffer2[10 + 1];
- snprintfz(buffer2, 10, "%d", default_port);
-
- char *ip = buffer, *port = buffer2;
-
- char *e = ip;
- if(*e == '[') {
- e = ++ip;
- while(*e && *e != ']') e++;
- if(*e == ']') {
- *e = '\0';
- e++;
- }
- }
- else {
- while(*e && *e != ':') e++;
- }
-
- if(*e == ':') {
- port = e + 1;
- *e = '\0';
- }
-
- if(!*ip)
- return -1;
-
- if(!*port)
- port = buffer2;
-
- memset(&hints, 0, sizeof(struct addrinfo));
- hints.ai_family = AF_UNSPEC; /* Allow IPv4 or IPv6 */
- hints.ai_socktype = SOCK_DGRAM; /* Datagram socket */
- hints.ai_flags = AI_PASSIVE; /* For wildcard IP address */
- hints.ai_protocol = 0; /* Any protocol */
- hints.ai_canonname = NULL;
- hints.ai_addr = NULL;
- hints.ai_next = NULL;
-
- int r = getaddrinfo(ip, port, &hints, &result);
- if (r != 0) {
- error("Cannot resolve host '%s', port '%s': %s\n", ip, port, gai_strerror(r));
- return -1;
- }
-
- int fd = -1;
- for (rp = result; rp != NULL && fd == -1; rp = rp->ai_next) {
- char rip[INET_ADDRSTRLEN + INET6_ADDRSTRLEN] = "INVALID";
- int rport;
-
- switch (rp->ai_addr->sa_family) {
- case AF_INET: {
- 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, timeout);
- break;
- }
-
- case AF_INET6: {
- 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, timeout);
- break;
- }
- }
- }
-
- freeaddrinfo(result);
-
- return fd;
-}
-
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);
- if(unlikely(before - after < st->update_every && after != after - after % st->update_every))
- // when st->update_every is bigger than the frequency we send data to backend
- // skip the iterations that are not aligned to the database
+ if(unlikely(before < first_t || after > last_t))
+ // the chart has not been updated in the wanted timeframe
return NAN;
// align the time-frame
return 0;
}
+static inline int process_graphite_response(BUFFER *b) {
+ char sample[1024];
+ const char *s = buffer_tostring(b);
+ char *d = sample, *e = &sample[sizeof(sample) - 1];
+
+ for(; *s && d < e ;s++) {
+ char c = *s;
+ if(unlikely(!isprint(c))) c = ' ';
+ *d++ = c;
+ }
+ *d = '\0';
+
+ info("Received %zu bytes from graphite backend. Ignoring them. Sample: '%s'", buffer_strlen(b), sample);
+ buffer_flush(b);
+ return 0;
+}
+
+static inline int process_opentsdb_response(BUFFER *b) {
+ char sample[1024];
+ const char *s = buffer_tostring(b);
+ char *d = sample, *e = &sample[sizeof(sample) - 1];
+
+ for(; *s && d < e ;s++) {
+ char c = *s;
+ if(unlikely(!isprint(c))) c = ' ';
+ *d++ = c;
+ }
+ *d = '\0';
+
+ info("Received %zu bytes from opentsdb backend. Ignoring them. Sample: '%s'", buffer_strlen(b), sample);
+ buffer_flush(b);
+ return 0;
+}
+
void *backends_main(void *ptr) {
- (void)ptr;
+ struct netdata_static_thread *static_thread = (struct netdata_static_thread *)ptr;
- BUFFER *b = buffer_create(1);
- 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);
+ BUFFER *b = buffer_create(1), *response = buffer_create(1);
+ int (*backend_request_formatter)(BUFFER *b, const char *prefix, RRDHOST *host, const char *hostname, RRDSET *st, RRDDIM *rd, time_t after, time_t before, uint32_t options) = NULL;
+ int (*backend_response_checker)(BUFFER *b) = NULL;
info("BACKEND thread created with task id %d", gettid());
int default_port = 0;
int sock = -1;
uint32_t options;
- int enabled = config_get_boolean("backend", "enable", 0);
+ 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");
if(!strcmp(type, "graphite") || !strcmp(type, "graphite:plaintext")) {
default_port = 2003;
if(options == BACKEND_SOURCE_DATA_AS_COLLECTED)
- formatter = format_dimension_collected_graphite_plaintext;
+ backend_request_formatter = format_dimension_collected_graphite_plaintext;
else
- formatter = format_dimension_stored_graphite_plaintext;
+ backend_request_formatter = format_dimension_stored_graphite_plaintext;
+
+ backend_response_checker = process_graphite_response;
}
else if(!strcmp(type, "opentsdb") || !strcmp(type, "opentsdb:telnet")) {
default_port = 4242;
if(options == BACKEND_SOURCE_DATA_AS_COLLECTED)
- formatter = format_dimension_collected_opentsdb_telnet;
+ backend_request_formatter = format_dimension_collected_opentsdb_telnet;
else
- formatter = format_dimension_stored_opentsdb_telnet;
+ backend_request_formatter = format_dimension_stored_opentsdb_telnet;
+
+ backend_response_checker = process_opentsdb_response;
}
else {
error("Unknown backend type '%s'", type);
goto cleanup;
}
+ if(backend_request_formatter == NULL || backend_response_checker == NULL) {
+ error("backend is misconfigured - disabling it.");
+ goto cleanup;
+ }
+
if(timeoutms < 1) {
error("BACKED invalid timeout %ld ms given. Assuming %d ms.", timeoutms, frequency * 2 * 1000);
timeoutms = frequency * 2 * 1000;
chart_lost_metrics = 0,
chart_sent_metrics = 0,
chart_buffered_bytes = 0,
+ chart_received_bytes = 0,
chart_sent_bytes = 0,
+ chart_receptions = 0,
chart_transmission_successes = 0,
chart_transmission_failures = 0,
chart_data_lost_events = 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);
+ 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);
+ 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);
+ rrddim_add(chart_bytes, "received", 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);
+ 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);
+ rrddim_add(chart_ops, "read", NULL, 1, 1, RRDDIM_ABSOLUTE);
}
+ /*
+ * this is misleading - we can only measure the time we need to send data
+ * this time is not related to the time required for the data to travel to
+ * the backend database and the time that server needed to process them
+ *
+ * issue #1432 and https://www.softlab.ntua.gr/facilities/documentation/unix/unix-socket-faq/unix-socket-faq-2.html
+ *
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) {
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);
- time_t after = before;
+ usec_t step_ut = frequency * USEC_PER_SEC;
+ time_t after = now_realtime_sec();
int failures = 0;
+ heartbeat_t hb;
+ heartbeat_init(&hb);
for(;;) {
// ------------------------------------------------------------------------
- // wait for the next iteration point
-
- unsigned long long now_ut = time_usec();
- unsigned long long next_ut = now_ut - (now_ut % step_ut) + step_ut;
- before = (time_t)(next_ut / 1000000ULL);
-
- // add a little delay (1/4 of the step) plus some randomness
- next_ut += (step_ut / 4) + random_ut;
-
- while(now_ut < next_ut) {
- sleep_usec(next_ut - now_ut);
- now_ut = time_usec();
- }
+ // Wait for the next iteration point.
+ heartbeat_next(&hb, step_ut);
+ time_t before = now_realtime_sec();
// ------------------------------------------------------------------------
// 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)
- chart_buffered_metrics += formatter(b, prefix, &localhost, hostname, st, rd, after, before, options);
+ chart_buffered_metrics += backend_request_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_received_bytes =
chart_sent_bytes =
chart_sent_metrics =
chart_lost_metrics =
//fprintf(stderr, "\nBACKEND BEGIN:\n%s\nBACKEND END\n", buffer_tostring(b)); // FIXME
//fprintf(stderr, "after = %lu, before = %lu\n", after, before);
+ // prepare for the next iteration
+ // to add incrementally data to buffer
+ after = before;
+
// ------------------------------------------------------------------------
- // connect to a backend server
+ // if we are connected, receive a response, without blocking
+
+ if(likely(sock != -1)) {
+ errno = 0;
+
+ // loop through to collect all data
+ while(sock != -1 && errno != EWOULDBLOCK) {
+ buffer_need_bytes(response, 4096);
+
+ ssize_t r = recv(sock, &response->buffer[response->len], response->size - response->len, MSG_DONTWAIT);
+ if(likely(r > 0)) {
+ // we received some data
+ response->len += r;
+ chart_received_bytes += r;
+ chart_receptions++;
+ }
+ else if(r == 0) {
+ error("Backend '%s' closed the socket", destination);
+ close(sock);
+ sock = -1;
+ }
+ else {
+ // failed to receive data
+ if(errno != EAGAIN && errno != EWOULDBLOCK) {
+ error("Cannot receive data from backend '%s'.", destination);
+ }
+ }
+ }
+
+ // if we received data, process them
+ if(buffer_strlen(response))
+ backend_response_checker(response);
+ }
+
+ // ------------------------------------------------------------------------
+ // if we are not connected, connect to a backend server
if(unlikely(sock == -1)) {
- unsigned long long start_ut = time_usec();
+ usec_t start_ut = now_monotonic_usec();
const char *s = destination;
while(*s) {
const char *e = s;
char buf[e - s + 1];
strncpyz(buf, s, e - s);
chart_backend_reconnects++;
- sock = connect_to_one(buf, default_port, &timeout);
+ sock = connect_to(buf, default_port, &timeout);
if(sock != -1) break;
s = e;
}
- chart_backend_latency += time_usec() - start_ut;
+ chart_backend_latency += now_monotonic_usec() - start_ut;
}
if(unlikely(netdata_exit)) break;
// ------------------------------------------------------------------------
- // send our buffer to the backend server
+ // if we are connected, send our buffer to the backend server
if(likely(sock != -1)) {
size_t len = buffer_strlen(b);
- unsigned long long start_ut = time_usec();
+ usec_t start_ut = now_monotonic_usec();
int flags = 0;
#ifdef MSG_NOSIGNAL
flags += MSG_NOSIGNAL;
#endif
+
ssize_t written = send(sock, buffer_tostring(b), len, flags);
- chart_backend_latency += time_usec() - start_ut;
+ chart_backend_latency += now_monotonic_usec() - start_ut;
if(written != -1 && (size_t)written == len) {
// we sent the data successfully
chart_transmission_successes++;
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 {
error("Failed to update database backend '%s'", destination);
// update the monitoring charts
if(chart_ops->counter_done) rrdset_next(chart_ops);
+ rrddim_set(chart_ops, "read", chart_receptions);
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_bytes, "buffered", chart_buffered_bytes);
rrddim_set(chart_bytes, "lost", chart_lost_bytes);
rrddim_set(chart_bytes, "sent", chart_sent_bytes);
+ rrddim_set(chart_bytes, "received", chart_received_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);
if(sock != -1)
close(sock);
+ buffer_free(b);
+ buffer_free(response);
+
info("BACKEND thread exiting");
+ static_thread->enabled = 0;
pthread_exit(NULL);
return NULL;
}