purgeService->getHealthEndpoint(); if ($endpoint === '') { $output->writeln('SSR_HEALTH_ENDPOINT missing in var/vitec-ssr-purge.env'); return Command::INVALID; } $row = ['crdate' => time(), 'ok' => 0, 'release_id' => '', 'cached_pages' => 0, 'cache_mb' => 0, 'response_ms' => 0]; $start = microtime(true); try { $response = $this->requestFactory->request($endpoint, 'GET', [ 'timeout' => 4, 'connect_timeout' => 2, 'http_errors' => false, ]); $row['response_ms'] = (int)round((microtime(true) - $start) * 1000); if ($response->getStatusCode() === 200) { $data = json_decode((string)$response->getBody(), true); $row['ok'] = !empty($data['ok']) ? 1 : 0; $row['release_id'] = substr((string)($data['release']['release'] ?? ''), 0, 100); $row['cached_pages'] = (int)($data['cachedPages'] ?? 0); $row['cache_mb'] = (int)($data['cacheMB'] ?? 0); } } catch (\Throwable $e) { $row['response_ms'] = (int)round((microtime(true) - $start) * 1000); } // Cumulative node counters from /__stats (-1 = no data this sample, // e.g. node down or release without the endpoint). The errors tile // computes 24h deltas from these, skipping -1 rows; the counters // reset whenever the node restarts, which the delta logic clamps. $stats = [ 'stat_hit' => -1, 'stat_miss' => -1, 'stat_stale' => -1, 'stat_bypass' => -1, 'stat_render_errors' => -1, 'stat_backend_errors' => -1, ]; if ($row['ok'] === 1 && ($statsEndpoint = $this->purgeService->getStatsEndpoint()) !== '') { try { $statsResponse = $this->requestFactory->request($statsEndpoint, 'GET', [ 'timeout' => 4, 'connect_timeout' => 2, 'http_errors' => false, ]); if ($statsResponse->getStatusCode() === 200) { $s = json_decode((string)$statsResponse->getBody(), true) ?: []; $stats = [ 'stat_hit' => (int)($s['hit'] ?? -1), 'stat_miss' => (int)($s['miss'] ?? -1), 'stat_stale' => (int)($s['stale'] ?? -1), 'stat_bypass' => (int)($s['bypass'] ?? -1), 'stat_render_errors' => (int)($s['renderErrors'] ?? -1), 'stat_backend_errors' => (int)($s['backendErrors'] ?? -1), ]; } } catch (\Throwable) { // Stats are optional - the health sample must never fail on them. } } $row = array_merge($row, $stats); // Previous samples, newest first, read BEFORE this one is inserted: // [0] is the state every transition decision compares against. $failoverAfter = $this->purgeService->getFailoverAfter(); $previous = array_map('intval', $this->connectionPool ->getQueryBuilderForTable('tx_vitec_ssrhealthlog') ->select('ok')->from('tx_vitec_ssrhealthlog') ->orderBy('crdate', 'DESC')->addOrderBy('uid', 'DESC') ->setMaxResults($failoverAfter) ->executeQuery()->fetchFirstColumn()); $conn = $this->connectionPool->getConnectionForTable('tx_vitec_ssrhealthlog'); $conn->insert('tx_vitec_ssrhealthlog', $row); $conn->executeStatement( 'DELETE FROM tx_vitec_ssrhealthlog WHERE crdate < ?', [time() - self::KEEP_DAYS * 86400] ); $output->writeln(sprintf( '%s release=%s pages=%d cache=%dMB %dms', $row['ok'] ? 'ok' : 'DOWN', $row['release_id'] ?: '-', $row['cached_pages'], $row['cache_mb'], $row['response_ms'] )); if ($row['stat_hit'] >= 0) { $output->writeln(sprintf( 'stats: hit=%d miss=%d stale=%d bypass=%d renderErrors=%d backendErrors=%d', $row['stat_hit'], $row['stat_miss'], $row['stat_stale'], $row['stat_bypass'], $row['stat_render_errors'], $row['stat_backend_errors'] )); } $prevOk = $previous === [] ? null : $previous[0]; if ($prevOk !== null && $prevOk !== $row['ok']) { if ($row['ok'] === 1) { $this->sendAlert( '[VITEC SSR] Entwarnung: SSR-Server wieder erreichbar', "Die 5-Minuten-Messung erreicht den SSR-Server wieder (Status ok).\n\n" . 'Zeitpunkt: ' . date('d.m.Y H:i:s') . "\n" . 'Release: ' . ($row['release_id'] ?: '-') . "\n" . 'Antwortzeit: ' . $row['response_ms'] . " ms\n\n" . 'Aktueller Stand: Schalter vitec.ssr.enabled = ' . ($this->flagService->isEnabledInSettings() ? 'AN' : 'AUS') . ', Flag-Datei ' . ($this->flagService->isFlagSet() ? 'vorhanden (SSR aktiv)' : 'fehlt (SPA-Auslieferung)') . ".\n" . "Falls der Auto-Failover ausgelöst wurde, bleibt die Seite bewusst auf SPA.\n" . "Wieder einschalten: TYPO3-Backend > Site Management > Settings > SSR delivery auf AN,\n" . "danach Komplett-Purge (vendor/bin/typo3 vitec:ssr-purge --all).\n", $output ); } else { $this->sendAlert( '[VITEC SSR] ALARM: SSR-Server antwortet nicht', "Die 5-Minuten-Messung konnte den SSR-Server nicht erreichen\n" . "(oder /__health meldete einen Fehler).\n\n" . 'Zeitpunkt: ' . date('d.m.Y H:i:s') . "\n" . 'Endpoint: ' . $endpoint . "\n" . 'Antwortzeit: ' . $row['response_ms'] . " ms\n\n" . 'Solange die Flag-Datei existiert, leitet dev.vitec.com weiter zum SSR-Server. Nach ' . $failoverAfter . ' Fehlmessungen in Folge (ca. ' . ($failoverAfter * 5) . " Minuten) schaltet\n" . "der Auto-Failover automatisch auf SPA-Auslieferung um - die Seite bleibt erreichbar.\n\n" . "Diese Mail kommt nur beim Zustandswechsel ok -> down, nicht bei jeder Messung.\n", $output ); } } if ($row['ok'] === 0 && $this->flagService->isFlagSet()) { $consecutive = 1; foreach ($previous as $ok) { if ($ok !== 0) { break; } $consecutive++; } if ($consecutive >= $failoverAfter) { try { $this->flagService->disableSsr(); $output->writeln(sprintf( 'AUTO-FAILOVER after %d consecutive misses: vitec.ssr.enabled=false, caches flushed, flag removed', $consecutive )); $this->sendAlert( '[VITEC SSR] AUTO-FAILOVER: Seite auf SPA umgeschaltet', 'AUTO-FAILOVER AUSGELÖST: ' . $consecutive . " Messungen in Folge down.\n\n" . "dev.vitec.com liefert ab sofort die SPA aus (Rendering im Browser).\n" . "Für Besucher bleibt die Seite erreichbar.\n\n" . "Was automatisch passiert ist:\n" . "- Site-Setting vitec.ssr.enabled auf false gestellt (settings.yaml)\n" . "- Caches geleert\n" . "- Flag-Datei var/vitec-ssr-on.flag gelöscht\n\n" . "Wieder einschalten (bewusst NUR von Hand), sobald der SSR-Server wieder läuft:\n" . "1. Auf dem VPS prüfen: systemctl status vitec-ssr und /__health\n" . "2. TYPO3-Backend > Site Management > Settings > SSR delivery wieder AN\n" . "3. Komplett-Purge: vendor/bin/typo3 vitec:ssr-purge --all\n", $output ); } catch (\Throwable $e) { $output->writeln('auto-failover FAILED: ' . $e->getMessage() . ''); $this->sendAlert( '[VITEC SSR] FEHLER: Auto-Failover fehlgeschlagen', "Der Auto-Failover sollte auslösen, ist aber fehlgeschlagen:\n" . $e->getMessage() . "\n\n" . "Bitte von Hand prüfen: Site-Setting vitec.ssr.enabled, Datei var/vitec-ssr-on.flag.\n" . "Not-Aus ohne TYPO3: Flag-Datei löschen UND Setting auf false stellen\n" . "(sonst legt die SyncSsrFlag-Middleware die Flag-Datei sofort wieder an).\n", $output ); } } } return Command::SUCCESS; } /** * Mails are best effort: a mail problem must never break the sampler * (the cron would stop logging history). Recipients come from * SSR_ALERT_EMAIL in var/vitec-ssr-purge.env - unset means "alerts * off", which the output states so the cron log shows why. */ private function sendAlert(string $subject, string $body, OutputInterface $output): void { $recipients = GeneralUtility::trimExplode(',', $this->purgeService->getAlertEmail(), true); if ($recipients === []) { $output->writeln('transition detected but SSR_ALERT_EMAIL is not configured - no mail sent'); return; } try { $mail = GeneralUtility::makeInstance(MailMessage::class); $mail->to(...$recipients)->subject($subject)->text($body); GeneralUtility::makeInstance(Mailer::class)->send($mail); $output->writeln('alert mail sent: ' . $subject . ' -> ' . implode(', ', $recipients)); } catch (\Throwable $e) { $output->writeln('alert mail failed: ' . $e->getMessage() . ''); } } }