anuragaw commented on a change in pull request #3575: [WIP DO NOT MERGE] Health check feature for virtual router URL: https://github.com/apache/cloudstack/pull/3575#discussion_r358788702
########## File path: server/src/main/java/com/cloud/network/router/VirtualNetworkApplianceManagerImpl.java ########## @@ -1186,6 +1227,431 @@ protected void pushToUpdateQueue(final List<NetworkVO> networks) throws Interrup } } + protected class FetchRouterHealthChecksResultTask extends ManagedContextRunnable { + public FetchRouterHealthChecksResultTask() { + } + + @Override + protected void runInContext() { + try { + final List<DomainRouterVO> routers = _routerDao.listByStateAndManagementServer(VirtualMachine.State.Running, mgmtSrvrId); + s_logger.debug("Found " + routers.size() + " running routers. "); + + for (final DomainRouterVO router : routers) { + GetRouterMonitorResultsAnswer answer = fetchAndUpdateRouterHealthChecks(router, false); + String checkFailsToRestartVr = RouterHealthChecksFailuresToRestartVr.valueIn(router.getDataCenterId()); + if (answer == null) { + s_logger.warn("Unable to fetch monitor results for router " + router); + updateRouterConnectivityHealthCheck(router.getId(), false, "Communication failed"); + } else if (!answer.getResult()) { + s_logger.warn("Failed to fetch monitor results from router " + router + " with details: " + answer.getDetails()); + updateRouterConnectivityHealthCheck(router.getId(), false, "Failed to fetch results with details: " + answer.getDetails()); + } else { + updateRouterConnectivityHealthCheck(router.getId(), true, "Successfully fetched data"); + parseAndMergeDbHealthChecks(router.getId(), answer.getMonitoringResults()); + + // Check failing tests and restart if needed + if (answer.getFailingChecks().size() > 0 && StringUtils.isNotBlank(checkFailsToRestartVr)) { + s_logger.info("Checking failed health checks to see if router needs reboot"); + for (String failedCheck : answer.getFailingChecks()) { + if (checkFailsToRestartVr.contains(failedCheck)) { + s_logger.info("Found failing health check " + failedCheck + " so restarting router."); + rebootRouter(router.getId(), true); + } + } + } + } + } + } catch (final Exception ex) { + s_logger.error("Fail to complete the FetchRouterHealthChecksResultTask! ", ex); + ex.printStackTrace(); + } + } + } + + private Map<String, Map<String, RouterHealthCheckResultVO>> getHealthChecksFromDb(long routerId) { + List<RouterHealthCheckResultVO> healthChecksList = routerHealthCheckResultDao.getHealthCheckResults(routerId); + Map<String, Map<String, RouterHealthCheckResultVO>> healthCheckResults = new HashMap<>(); + if (healthChecksList.isEmpty()) { + return healthCheckResults; + } + + for (RouterHealthCheckResultVO healthCheck : healthChecksList) { + if (!healthCheckResults.containsKey(healthCheck.getCheckType())) { + healthCheckResults.put(healthCheck.getCheckType(), new HashMap<>()); + } + healthCheckResults.get(healthCheck.getCheckType()).put(healthCheck.getCheckName(), healthCheck); + } + + return healthCheckResults; + } + + private RouterHealthCheckResultVO updateRouterConnectivityHealthCheck(final long routerId, boolean connected, String message) { + boolean newEntry = false; + RouterHealthCheckResultVO connectivityVO = routerHealthCheckResultDao.getRouterHealthCheckResult(routerId, "connectivity", "basic"); + if (connectivityVO == null) { + connectivityVO = new RouterHealthCheckResultVO(routerId, "connectivity", "basic"); + newEntry = true; + } + + connectivityVO.setCheckResult(connected); + connectivityVO.setLastUpdateTime(new Date()); + connectivityVO.setCheckDetails(StringUtils.isNotEmpty(message) ? message.getBytes(Charset.forName("US-ASCII")) : null); + + if (newEntry) { + routerHealthCheckResultDao.persist(connectivityVO); + } else { + routerHealthCheckResultDao.update(connectivityVO.getId(), connectivityVO); + } + + return routerHealthCheckResultDao.getRouterHealthCheckResult(routerId, "connectivity", "basic"); + } + + private List<RouterHealthCheckResult> parseAndMergeDbHealthChecks(final long routerId, final String monitoringResult) { + if (StringUtils.isBlank(monitoringResult)) { + s_logger.warn("Attempted parsing empty monitoring results string for router " + routerId); + return Collections.emptyList(); + } + + try { + s_logger.info("Parsing and updating DB health check data for router: " + routerId); + s_logger.info("Retrieved data" + monitoringResult); + final Map<String, Map<String, RouterHealthCheckResultVO>> checksInDb = getHealthChecksFromDb(routerId); + final Type t = new TypeToken<Map<String, Map<String, Map<String, String>>>>() {}.getType(); + final Map<String, Map<String, Map<String, String>>> checks = GsonHelper.getGson().fromJson(monitoringResult, t); + final String lastRunKey = "lastRun"; + List<RouterHealthCheckResult> healthChecks = new ArrayList<>(); + for (String checkType : checks.keySet()) { + if (checks.get(checkType).containsKey(lastRunKey)) { + Map<String, String> lastRun = checks.get(checkType).get(lastRunKey); + s_logger.info("Found check types executed on VR " + checkType + ", start: " + lastRun.get("start") + + ", end: " + lastRun.get("end") + ", duration: " + lastRun.get("duration")); + } + + for (String checkName : checks.get(checkType).keySet()) { + if (lastRunKey.equals(checkName)) { + continue; + } + Map<String, String> checkData = checks.get(checkType).get(checkName); + try { + boolean success = Boolean.parseBoolean(checkData.get("success")); + Date lastUpdate = new Date(Long.parseLong(checkData.get("lastUpdate"))); + double lastRunDuration = Double.parseDouble(checkData.get("lastRunDuration")); + String message = checkData.get("message"); + final RouterHealthCheckResultVO hcVo; + boolean newEntry = false; + if (checksInDb.containsKey(checkType) && checksInDb.get(checkType).containsKey(checkName)) { + hcVo = checksInDb.get(checkType).get(checkName); + } else { + hcVo = new RouterHealthCheckResultVO(routerId, checkName, checkType); + newEntry = true; + } + + hcVo.setCheckResult(success); + hcVo.setLastUpdateTime(lastUpdate); + hcVo.setCheckDetails(StringUtils.isNotEmpty(message) ? message.getBytes(Charset.forName("US-ASCII")) : null); + if (newEntry) { + routerHealthCheckResultDao.persist(hcVo); + } else { + routerHealthCheckResultDao.update(hcVo.getId(), hcVo); + } + healthChecks.add(hcVo); + s_logger.info("Found health check " + hcVo + " which took running duration (secs) " + lastRunDuration); + } catch (Exception ex) { + s_logger.error("Skipping health check: Exception while parsing check result data for router id " + routerId + + ", check type: " + checkType + ", check name: " + checkName + ":" + ex.getLocalizedMessage()); + ex.printStackTrace(); + } + } + } + return healthChecks; + } catch (JsonSyntaxException ex) { + s_logger.error("Unable to parse the result of health checks due to " + ex.getLocalizedMessage()); + ex.printStackTrace(); + } + + return Collections.emptyList(); + } + + // Returns null if health checks are not enabled Review comment: removed it, it was left over on private method. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services