diff options
Diffstat (limited to 'src/backends.c')
| -rw-r--r-- | src/backends.c | 549 |
1 files changed, 549 insertions, 0 deletions
diff --git a/src/backends.c b/src/backends.c new file mode 100644 index 000000000..1272d0473 --- /dev/null +++ b/src/backends.c @@ -0,0 +1,549 @@ +#include "common.h" + +#define BACKEND_SOURCE_DATA_AS_COLLECTED 0x00000001 +#define BACKEND_SOURCE_DATA_AVERAGE 0x00000002 +#define BACKEND_SOURCE_DATA_SUM 0x00000004 + +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 + return NAN; + + // align the time-frame + // for 'after' also skip the first value by adding st->update_every + after = after - after % st->update_every + st->update_every; + before = before - before % st->update_every; + + if(unlikely(after < first_t)) + after = first_t; + + if(unlikely(after > before)) + // this can happen when the st->update_every > before - after + before = after; + + if(unlikely(before > last_t)) + before = last_t; + + size_t counter = 0; + calculated_number sum = 0; + + long start_at_slot = rrdset_time2slot(st, before), + stop_at_slot = rrdset_time2slot(st, after), + slot, stop_now = 0; + + for(slot = start_at_slot; !stop_now ; slot--) { + if(unlikely(slot < 0)) slot = st->entries - 1; + if(unlikely(slot == stop_at_slot)) stop_now = 1; + + storage_number n = rd->values[slot]; + if(unlikely(!does_storage_number_exist(n))) continue; + + calculated_number value = unpack_storage_number(n); + sum += value; + counter++; + } + + if(unlikely(!counter)) + return NAN; + + if(unlikely(options & BACKEND_SOURCE_DATA_SUM)) + return sum; + + return sum / (calculated_number)counter; +} + +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 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_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 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 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_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; +} + +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) { + struct netdata_static_thread *static_thread = (struct netdata_static_thread *)ptr; + + 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()); + + if(pthread_setcanceltype(PTHREAD_CANCEL_DEFERRED, NULL) != 0) + error("Cannot set pthread cancel type to DEFERRED."); + + 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; + 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", 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; + + if(!strcmp(source, "as collected")) { + options = BACKEND_SOURCE_DATA_AS_COLLECTED; + } + else if(!strcmp(source, "average")) { + options = BACKEND_SOURCE_DATA_AVERAGE; + } + else if(!strcmp(source, "sum") || !strcmp(source, "volume")) { + options = BACKEND_SOURCE_DATA_SUM; + } + else { + error("Invalid data source method '%s' for backend given. Disabling backed.", source); + goto cleanup; + } + + if(!strcmp(type, "graphite") || !strcmp(type, "graphite:plaintext")) { + default_port = 2003; + if(options == BACKEND_SOURCE_DATA_AS_COLLECTED) + backend_request_formatter = format_dimension_collected_graphite_plaintext; + else + 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) + backend_request_formatter = format_dimension_collected_opentsdb_telnet; + else + 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; + } + 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_received_bytes = 0, + chart_sent_bytes = 0, + chart_receptions = 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); + 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, "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) { + 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); + + usec_t step_ut = frequency * USEC_PER_SEC; + usec_t random_ut = now_realtime_usec() % (step_ut / 2); + time_t before = (time_t)((now_realtime_usec() - step_ut) / USEC_PER_SEC); + time_t after = before; + int failures = 0; + + for(;;) { + // ------------------------------------------------------------------------ + // wait for the next iteration point + + usec_t now_ut = now_realtime_usec(); + usec_t next_ut = now_ut - (now_ut % step_ut) + step_ut; + before = (time_t)(next_ut / USEC_PER_SEC); + + // 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 = now_realtime_usec(); + } + + // ------------------------------------------------------------------------ + // add to the buffer the data we need to send to the backend + + RRDSET *st; + int pthreadoldcancelstate; + + if(unlikely(pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &pthreadoldcancelstate) != 0)) + error("Cannot set pthread cancel state to DISABLE."); + + rrdhost_rdlock(&localhost); + for(st = localhost.rrdset_root; st ;st = st->next) { + pthread_rwlock_rdlock(&st->rwlock); + + RRDDIM *rd; + for(rd = st->dimensions; rd ;rd = rd->next) { + if(rd->last_collected_time.tv_sec >= after) + chart_buffered_metrics += backend_request_formatter(b, prefix, &localhost, hostname, st, rd, after, before, options); + } + + pthread_rwlock_unlock(&st->rwlock); + } + rrdhost_unlock(&localhost); + + 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 = + 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)); // FIXME + //fprintf(stderr, "after = %lu, before = %lu\n", after, before); + + // ------------------------------------------------------------------------ + // 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)) { + usec_t start_ut = now_realtime_usec(); + const char *s = destination; + while(*s) { + const char *e = s; + + // skip separators, moving both s(tart) and e(nd) + while(isspace(*e) || *e == ',') s = ++e; + + // move e(nd) to the first separator + while(*e && !isspace(*e) && *e != ',') e++; + + // is there anything? + if(!*s || s == e) break; + + char buf[e - s + 1]; + strncpyz(buf, s, e - s); + chart_backend_reconnects++; + sock = connect_to(buf, default_port, &timeout); + if(sock != -1) break; + s = e; + } + chart_backend_latency += now_realtime_usec() - start_ut; + } + + if(unlikely(netdata_exit)) break; + + // ------------------------------------------------------------------------ + // if we are connected, send our buffer to the backend server + + if(likely(sock != -1)) { + size_t len = buffer_strlen(b); + usec_t 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 { + // 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 { + 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, "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_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); + 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); + 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: + if(sock != -1) + close(sock); + + buffer_free(b); + buffer_free(response); + + info("BACKEND thread exiting"); + + static_thread->enabled = 0; + pthread_exit(NULL); + return NULL; +} |