public class DownloadProducerFromCloudsoftRepo
implements Function<DownloadRequirement, DownloadTargets> {
@SuppressWarnings("unused")
private static final Logger LOG =
LoggerFactory.getLogger(DownloadProducerFromCloudsoftRepo.class);
public static final ConfigKey<String> CLOUDSOFT_REPO_URL =
BasicConfigKey.builder(String.class)
.name(DownloadProducerFromProperties.DOWNLOAD_CONF_PREFIX + "repo.cloudsoft.url")
.description(
"Whether to use the cloudsoft repo for downloading entities, during installs")
.defaultValue("http://downloads.cloudsoftcorp.com/brooklyn/repository")
.build();
public static final ConfigKey<Boolean> CLOUDSOFT_REPO_ENABLED =
BasicConfigKey.builder(Boolean.class)
.name(DownloadProducerFromProperties.DOWNLOAD_CONF_PREFIX + "repo.cloudsoft.enabled")
.description(
"Whether to use the cloudsoft repo for downloading entities, during installs")
.defaultValue(true)
.build();
public static final String CLOUDSOFT_REPO_URL_PATTERN =
"%s/"
+ "${simpletype}/${version}/"
+ "<#if filename??>"
+ "${filename}"
+ "<#else>"
+ "<#if addon??>"
+ "${simpletype?lower_case}-${addon?lower_case}-${addonversion?lower_case}.${fileSuffix!\"tar.gz\"}"
+ "<#else>"
+ "${simpletype?lower_case}-${version?lower_case}.${fileSuffix!\"tar.gz\"}"
+ "</#if>"
+ "</#if>";
private final StringConfigMap config;
public DownloadProducerFromCloudsoftRepo(StringConfigMap config) {
this.config = config;
}
public DownloadTargets apply(DownloadRequirement req) {
Boolean enabled = config.getConfig(CLOUDSOFT_REPO_ENABLED);
String baseUrl = config.getConfig(CLOUDSOFT_REPO_URL);
String url = String.format(CLOUDSOFT_REPO_URL_PATTERN, baseUrl);
if (enabled) {
Map<String, ?> subs = DownloadSubstituters.getBasicSubstitutions(req);
String result = DownloadSubstituters.substitute(url, subs);
return BasicDownloadTargets.builder().addPrimary(result).build();
} else {
return BasicDownloadTargets.empty();
}
}
}
// @Catalog(name="Service Failure Detector", description="HA policy for deteting failure of a
// service")
public class ServiceFailureDetector extends ServiceStateLogic.ComputeServiceState {
// TODO Remove duplication between this and MemberFailureDetectionPolicy.
// The latter could be re-written to use this. Or could even be deprecated
// in favour of this.
public enum LastPublished {
NONE,
FAILED,
RECOVERED;
}
private static final Logger LOG = LoggerFactory.getLogger(ServiceFailureDetector.class);
private static final long MIN_PERIOD_BETWEEN_EXECS_MILLIS = 100;
public static final BasicNotificationSensor<FailureDescriptor> ENTITY_FAILED =
HASensors.ENTITY_FAILED;
@SetFromFlag("onlyReportIfPreviouslyUp")
public static final ConfigKey<Boolean> ENTITY_FAILED_ONLY_IF_PREVIOUSLY_UP =
ConfigKeys.newBooleanConfigKey(
"onlyReportIfPreviouslyUp",
"Prevents the policy from emitting ENTITY_FAILED if the entity fails on startup (ie has never been up)",
true);
public static final ConfigKey<Boolean> MONITOR_SERVICE_PROBLEMS =
ConfigKeys.newBooleanConfigKey(
"monitorServiceProblems",
"Whether to monitor service problems, and emit on failures there (if set to false, this monitors only service up)",
true);
@SetFromFlag("serviceOnFireStabilizationDelay")
public static final ConfigKey<Duration> SERVICE_ON_FIRE_STABILIZATION_DELAY =
BasicConfigKey.builder(Duration.class)
.name("serviceOnFire.stabilizationDelay")
.description(
"Time period for which the service must be consistently down for (e.g. doesn't report down-up-down) before concluding ON_FIRE")
.defaultValue(Duration.ZERO)
.build();
@SetFromFlag("entityFailedStabilizationDelay")
public static final ConfigKey<Duration> ENTITY_FAILED_STABILIZATION_DELAY =
BasicConfigKey.builder(Duration.class)
.name("entityFailed.stabilizationDelay")
.description(
"Time period for which the service must be consistently down for (e.g. doesn't report down-up-down) before emitting ENTITY_FAILED")
.defaultValue(Duration.ZERO)
.build();
@SetFromFlag("entityRecoveredStabilizationDelay")
public static final ConfigKey<Duration> ENTITY_RECOVERED_STABILIZATION_DELAY =
BasicConfigKey.builder(Duration.class)
.name("entityRecovered.stabilizationDelay")
.description(
"For a failed entity, time period for which the service must be consistently up for (e.g. doesn't report up-down-up) before emitting ENTITY_RECOVERED")
.defaultValue(Duration.ZERO)
.build();
protected Long firstUpTime;
protected Long currentFailureStartTime = null;
protected Long currentRecoveryStartTime = null;
protected Long publishEntityFailedTime = null;
protected Long publishEntityRecoveredTime = null;
protected Long setEntityOnFireTime = null;
protected LastPublished lastPublished = LastPublished.NONE;
private final AtomicBoolean executorQueued = new AtomicBoolean(false);
private volatile long executorTime = 0;
/**
* TODO Really don't want this mutex! ServiceStateLogic.setExpectedState() will call into
* `onEvent(null)`, so could get concurrent calls. How to handle that? I don't think
* `ServiceStateLogic.setExpectedState` should be making the call, but presumably that is their to
* remove a race condition so it is set before method returns. Caller shouldn't rely on that
* though. e.g. see `ServiceFailureDetectorTest.testNotifiedOfFailureOnStateOnFire`, where we get
* two notifications.
*/
private final Object mutex = new Object();
public ServiceFailureDetector() {
this(new ConfigBag());
}
public ServiceFailureDetector(Map<String, ?> flags) {
this(new ConfigBag().putAll(flags));
}
public ServiceFailureDetector(ConfigBag configBag) {
// TODO hierarchy should use ConfigBag, and not change flags
super(configBag.getAllConfigMutable());
}
@Override
public void onEvent(SensorEvent<Object> event) {
if (firstUpTime == null) {
if (event != null
&& Attributes.SERVICE_UP.equals(event.getSensor())
&& Boolean.TRUE.equals(event.getValue())) {
firstUpTime = event.getTimestamp();
} else if (event == null && Boolean.TRUE.equals(entity.getAttribute(Attributes.SERVICE_UP))) {
// If this enricher is registered after the entity is up, then we'll get a "synthetic"
// onEvent(null)
firstUpTime = System.currentTimeMillis();
}
}
super.onEvent(event);
}
@Override
protected void setActualState(Lifecycle state) {
long now = System.currentTimeMillis();
synchronized (mutex) {
if (state == Lifecycle.ON_FIRE) {
if (lastPublished == LastPublished.FAILED) {
if (currentRecoveryStartTime != null) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} health-check for {}, component was recovering, now failing: {}",
new Object[] {this, entity, getExplanation(state)});
currentRecoveryStartTime = null;
publishEntityRecoveredTime = null;
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} health-check for {}, component still failed: {}",
new Object[] {this, entity, getExplanation(state)});
}
} else {
if (firstUpTime == null && getConfig(ENTITY_FAILED_ONLY_IF_PREVIOUSLY_UP)) {
// suppress; won't publish
} else if (currentFailureStartTime == null) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} health-check for {}, component now failing: {}",
new Object[] {this, entity, getExplanation(state)});
currentFailureStartTime = now;
publishEntityFailedTime =
currentFailureStartTime
+ getConfig(ENTITY_FAILED_STABILIZATION_DELAY).toMilliseconds();
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} health-check for {}, component continuing failing: {}",
new Object[] {this, entity, getExplanation(state)});
}
}
if (setEntityOnFireTime == null) {
setEntityOnFireTime =
now + getConfig(SERVICE_ON_FIRE_STABILIZATION_DELAY).toMilliseconds();
}
currentRecoveryStartTime = null;
publishEntityRecoveredTime = null;
} else if (state == Lifecycle.RUNNING) {
if (lastPublished == LastPublished.FAILED) {
if (currentRecoveryStartTime == null) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} health-check for {}, component now recovering: {}",
new Object[] {this, entity, getExplanation(state)});
currentRecoveryStartTime = now;
publishEntityRecoveredTime =
currentRecoveryStartTime
+ getConfig(ENTITY_RECOVERED_STABILIZATION_DELAY).toMilliseconds();
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} health-check for {}, component continuing recovering: {}",
new Object[] {this, entity, getExplanation(state)});
}
} else {
if (currentFailureStartTime != null) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} health-check for {}, component was failing, now healthy: {}",
new Object[] {this, entity, getExplanation(state)});
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} health-check for {}, component still healthy: {}",
new Object[] {this, entity, getExplanation(state)});
}
}
currentFailureStartTime = null;
publishEntityFailedTime = null;
setEntityOnFireTime = null;
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} health-check for {}, in unconfirmed sate: {}",
new Object[] {this, entity, getExplanation(state)});
}
long recomputeIn = Long.MAX_VALUE; // For whether to call recomputeAfterDelay
if (publishEntityFailedTime != null) {
long delayBeforeCheck = publishEntityFailedTime - now;
if (delayBeforeCheck <= 0) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} publishing failed (state={}; currentFailureStartTime={}; now={}",
new Object[] {
this,
state,
Time.makeDateString(currentFailureStartTime),
Time.makeDateString(now)
});
publishEntityFailedTime = null;
lastPublished = LastPublished.FAILED;
entity.emit(
HASensors.ENTITY_FAILED,
new HASensors.FailureDescriptor(entity, getFailureDescription(now)));
} else {
recomputeIn = Math.min(recomputeIn, delayBeforeCheck);
}
} else if (publishEntityRecoveredTime != null) {
long delayBeforeCheck = publishEntityRecoveredTime - now;
if (delayBeforeCheck <= 0) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} publishing recovered (state={}; currentRecoveryStartTime={}; now={}",
new Object[] {
this,
state,
Time.makeDateString(currentRecoveryStartTime),
Time.makeDateString(now)
});
publishEntityRecoveredTime = null;
lastPublished = LastPublished.RECOVERED;
entity.emit(HASensors.ENTITY_RECOVERED, new HASensors.FailureDescriptor(entity, null));
} else {
recomputeIn = Math.min(recomputeIn, delayBeforeCheck);
}
}
if (setEntityOnFireTime != null) {
long delayBeforeCheck = setEntityOnFireTime - now;
if (delayBeforeCheck <= 0) {
if (LOG.isDebugEnabled())
LOG.debug(
"{} setting on-fire, now that deferred period has passed (state={})",
new Object[] {this, state});
setEntityOnFireTime = null;
super.setActualState(state);
} else {
recomputeIn = Math.min(recomputeIn, delayBeforeCheck);
}
} else {
super.setActualState(state);
}
if (recomputeIn < Long.MAX_VALUE) {
recomputeAfterDelay(recomputeIn);
}
}
}
protected String getExplanation(Lifecycle state) {
Duration serviceFailedStabilizationDelay = getConfig(ENTITY_FAILED_STABILIZATION_DELAY);
Duration serviceRecoveredStabilizationDelay = getConfig(ENTITY_RECOVERED_STABILIZATION_DELAY);
return String.format(
"location=%s; status=%s; lastPublished=%s; timeNow=%s; "
+ "currentFailurePeriod=%s; currentRecoveryPeriod=%s",
entity.getLocations(),
(state != null ? state : "<unreported>"),
lastPublished,
Time.makeDateString(System.currentTimeMillis()),
(currentFailureStartTime != null ? getTimeStringSince(currentFailureStartTime) : "<none>")
+ " (stabilization "
+ Time.makeTimeStringRounded(serviceFailedStabilizationDelay)
+ ")",
(currentRecoveryStartTime != null ? getTimeStringSince(currentRecoveryStartTime) : "<none>")
+ " (stabilization "
+ Time.makeTimeStringRounded(serviceRecoveredStabilizationDelay)
+ ")");
}
private String getFailureDescription(long now) {
String description = null;
Map<String, Object> serviceProblems = entity.getAttribute(Attributes.SERVICE_PROBLEMS);
if (serviceProblems != null && !serviceProblems.isEmpty()) {
Entry<String, Object> problem = serviceProblems.entrySet().iterator().next();
description = problem.getKey() + ": " + problem.getValue();
if (serviceProblems.size() > 1) {
description = serviceProblems.size() + " service problems, including " + description;
} else {
description = "service problem: " + description;
}
} else if (Boolean.FALSE.equals(entity.getAttribute(Attributes.SERVICE_UP))) {
description = "service not up";
} else {
description = "service failure detected";
}
if (publishEntityFailedTime != null
&& currentFailureStartTime != null
&& publishEntityFailedTime > currentFailureStartTime)
description =
" (stabilized for "
+ Duration.of(now - currentFailureStartTime, TimeUnit.MILLISECONDS)
+ ")";
return description;
}
@SuppressWarnings({"unchecked", "rawtypes"})
protected void recomputeAfterDelay(long delay) {
if (isRunning() && executorQueued.compareAndSet(false, true)) {
long now = System.currentTimeMillis();
delay = Math.max(0, Math.max(delay, (executorTime + MIN_PERIOD_BETWEEN_EXECS_MILLIS) - now));
if (LOG.isTraceEnabled()) LOG.trace("{} scheduling publish in {}ms", this, delay);
Runnable job =
new Runnable() {
@Override
public void run() {
try {
executorTime = System.currentTimeMillis();
executorQueued.set(false);
onEvent(null);
} catch (Exception e) {
if (isRunning()) {
LOG.error("Error in enricher " + this + ": " + e, e);
} else {
if (LOG.isDebugEnabled())
LOG.debug("Error in enricher " + this + " (but no longer running): " + e, e);
}
} catch (Throwable t) {
LOG.error("Error in enricher " + this + ": " + t, t);
throw Exceptions.propagate(t);
}
}
};
ScheduledTask task =
new ScheduledTask(
MutableMap.of("delay", Duration.of(delay, TimeUnit.MILLISECONDS)),
new BasicTask(job));
((EntityInternal) entity).getExecutionContext().submit(task);
}
}
private String getTimeStringSince(Long time) {
return time == null ? null : Time.makeTimeStringRounded(System.currentTimeMillis() - time);
}
}
/**
* Monitors a given {@link HostAndPort}, to emit HASensors.CONNECTION_FAILED and
* HASensors.CONNECTION_RECOVERED if the connection is lost/restored.
*/
@Catalog(
name = "Connection Failure Detector",
description =
"HA policy for monitoring a host:port, "
+ "emitting an event if the connection is lost/restored")
public class ConnectionFailureDetector extends AbstractPolicy {
// TODO Remove duplication from ServiceFailureDetector, particularly for the stabilisation delays.
public enum LastPublished {
NONE,
FAILED,
RECOVERED;
}
private static final Logger LOG = LoggerFactory.getLogger(ConnectionFailureDetector.class);
private static final long MIN_PERIOD_BETWEEN_EXECS_MILLIS = 100;
public static final ConfigKey<HostAndPort> ENDPOINT =
ConfigKeys.newConfigKey(HostAndPort.class, "connectionFailureDetector.endpoint");
public static final ConfigKey<Duration> POLL_PERIOD =
ConfigKeys.newConfigKey(
Duration.class, "connectionFailureDetector.pollPeriod", "", Duration.ONE_SECOND);
public static final BasicNotificationSensor<FailureDescriptor> CONNECTION_FAILED =
HASensors.CONNECTION_FAILED;
public static final BasicNotificationSensor<FailureDescriptor> CONNECTION_RECOVERED =
HASensors.CONNECTION_RECOVERED;
@SetFromFlag("connectionFailedStabilizationDelay")
public static final ConfigKey<Duration> CONNECTION_FAILED_STABILIZATION_DELAY =
BasicConfigKey.builder(Duration.class)
.name("connectionFailureDetector.serviceFailedStabilizationDelay")
.description(
"Time period for which the connection must be consistently down for "
+ "(e.g. doesn't report down-up-down) before concluding failure. "
+ "Note that long TCP timeouts mean there can be long (e.g. 70 second) "
+ "delays in noticing a connection refused condition.")
.defaultValue(Duration.ZERO)
.build();
@SetFromFlag("connectionRecoveredStabilizationDelay")
public static final ConfigKey<Duration> CONNECTION_RECOVERED_STABILIZATION_DELAY =
BasicConfigKey.builder(Duration.class)
.name("connectionFailureDetector.serviceRecoveredStabilizationDelay")
.description(
"For a failed connection, time period for which the connection must be consistently up for (e.g. doesn't report up-down-up) before concluding recovered")
.defaultValue(Duration.ZERO)
.build();
protected final AtomicReference<Long> connectionLastUp = new AtomicReference<Long>();
protected final AtomicReference<Long> connectionLastDown = new AtomicReference<Long>();
protected Long currentFailureStartTime = null;
protected Long currentRecoveryStartTime = null;
protected LastPublished lastPublished = LastPublished.NONE;
private final AtomicBoolean executorQueued = new AtomicBoolean(false);
private volatile long executorTime = 0;
private Callable<Task<?>> pollingTaskFactory;
private Task<?> scheduledTask;
public ConnectionFailureDetector() {}
@Override
public void init() {
getRequiredConfig(ENDPOINT); // just to confirm it's set, failing fast
pollingTaskFactory =
new Callable<Task<?>>() {
@Override
public Task<?> call() {
BasicTask<Void> task =
new BasicTask<Void>(
new Runnable() {
@Override
public void run() {
checkHealth();
}
});
BrooklynTaskTags.setTransient(task);
return task;
}
};
}
@Override
public void setEntity(EntityLocal entity) {
super.setEntity(entity);
if (isRunning()) {
doStartPolling();
}
}
@Override
public void suspend() {
scheduledTask.cancel(true);
super.suspend();
}
@Override
public void resume() {
currentFailureStartTime = null;
currentRecoveryStartTime = null;
lastPublished = LastPublished.NONE;
executorQueued.set(false);
executorTime = 0;
super.resume();
doStartPolling();
}
protected void doStartPolling() {
if (scheduledTask == null || scheduledTask.isDone()) {
ScheduledTask task =
new ScheduledTask(MutableMap.of("period", getConfig(POLL_PERIOD)), pollingTaskFactory);
scheduledTask = ((EntityInternal) entity).getExecutionContext().submit(task);
}
}
private Duration getConnectionFailedStabilizationDelay() {
return getConfig(CONNECTION_FAILED_STABILIZATION_DELAY);
}
private Duration getConnectionRecoveredStabilizationDelay() {
return getConfig(CONNECTION_RECOVERED_STABILIZATION_DELAY);
}
private synchronized void checkHealth() {
CalculatedStatus status = calculateStatus();
boolean connected = status.connected;
long now = System.currentTimeMillis();
if (connected) {
connectionLastUp.set(now);
if (lastPublished == LastPublished.FAILED) {
if (currentRecoveryStartTime == null) {
LOG.info(
"{} connectivity-check for {}, now recovering: {}",
new Object[] {this, entity, status.getDescription()});
currentRecoveryStartTime = now;
schedulePublish();
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} connectivity-check for {}, continuing recovering: {}",
new Object[] {this, entity, status.getDescription()});
}
} else {
if (currentFailureStartTime != null) {
LOG.info(
"{} connectivity-check for {}, now healthy: {}",
new Object[] {this, entity, status.getDescription()});
currentFailureStartTime = null;
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} connectivity-check for {}, still healthy: {}",
new Object[] {this, entity, status.getDescription()});
}
}
} else {
connectionLastDown.set(now);
if (lastPublished != LastPublished.FAILED) {
if (currentFailureStartTime == null) {
LOG.info(
"{} connectivity-check for {}, now failing: {}",
new Object[] {this, entity, status.getDescription()});
currentFailureStartTime = now;
schedulePublish();
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} connectivity-check for {}, continuing failing: {}",
new Object[] {this, entity, status.getDescription()});
}
} else {
if (currentRecoveryStartTime != null) {
LOG.info(
"{} connectivity-check for {}, now failing: {}",
new Object[] {this, entity, status.getDescription()});
currentRecoveryStartTime = null;
} else {
if (LOG.isTraceEnabled())
LOG.trace(
"{} connectivity-check for {}, still failed: {}",
new Object[] {this, entity, status.getDescription()});
}
}
}
}
protected CalculatedStatus calculateStatus() {
return new CalculatedStatus();
}
protected void schedulePublish() {
schedulePublish(0);
}
protected void schedulePublish(long delay) {
if (isRunning() && executorQueued.compareAndSet(false, true)) {
long now = System.currentTimeMillis();
delay = Math.max(0, Math.max(delay, (executorTime + MIN_PERIOD_BETWEEN_EXECS_MILLIS) - now));
if (LOG.isTraceEnabled()) LOG.trace("{} scheduling publish in {}ms", this, delay);
Runnable job =
new Runnable() {
@Override
public void run() {
try {
executorTime = System.currentTimeMillis();
executorQueued.set(false);
publishNow();
} catch (Exception e) {
if (isRunning()) {
LOG.error("Problem resizing: " + e, e);
} else {
if (LOG.isDebugEnabled())
LOG.debug("Problem resizing, but no longer running: " + e, e);
}
} catch (Throwable t) {
LOG.error("Problem in service-failure-detector: " + t, t);
throw Exceptions.propagate(t);
}
}
};
ScheduledTask task =
new ScheduledTask(
MutableMap.of("delay", Duration.of(delay, TimeUnit.MILLISECONDS)),
new BasicTask(job));
((EntityInternal) entity).getExecutionContext().submit(task);
}
}
private synchronized void publishNow() {
if (!isRunning()) return;
CalculatedStatus calculatedStatus = calculateStatus();
boolean connected = calculatedStatus.connected;
Long lastUpTime = connectionLastUp.get();
Long lastDownTime = connectionLastDown.get();
long serviceFailedStabilizationDelay = getConnectionFailedStabilizationDelay().toMilliseconds();
long serviceRecoveredStabilizationDelay =
getConnectionRecoveredStabilizationDelay().toMilliseconds();
long now = System.currentTimeMillis();
if (connected) {
if (lastPublished == LastPublished.FAILED) {
// only publish if consistently up for serviceRecoveredStabilizationDelay
long currentRecoveryPeriod = getTimeDiff(now, currentRecoveryStartTime);
long sinceLastDownPeriod = getTimeDiff(now, lastDownTime);
if (currentRecoveryPeriod > serviceRecoveredStabilizationDelay
&& sinceLastDownPeriod > serviceRecoveredStabilizationDelay) {
String description = calculatedStatus.getDescription();
LOG.warn(
"{} connectivity-check for {}, publishing recovered: {}",
new Object[] {this, entity, description});
entity.emit(CONNECTION_RECOVERED, new HASensors.FailureDescriptor(entity, description));
lastPublished = LastPublished.RECOVERED;
currentFailureStartTime = null;
} else {
long nextAttemptTime =
Math.max(
serviceRecoveredStabilizationDelay - currentRecoveryPeriod,
serviceRecoveredStabilizationDelay - sinceLastDownPeriod);
schedulePublish(nextAttemptTime);
}
}
} else {
if (lastPublished != LastPublished.FAILED) {
// only publish if consistently down for serviceFailedStabilizationDelay
long currentFailurePeriod = getTimeDiff(now, currentFailureStartTime);
long sinceLastUpPeriod = getTimeDiff(now, lastUpTime);
if (currentFailurePeriod > serviceFailedStabilizationDelay
&& sinceLastUpPeriod > serviceFailedStabilizationDelay) {
String description = calculatedStatus.getDescription();
LOG.warn(
"{} connectivity-check for {}, publishing failed: {}",
new Object[] {this, entity, description});
entity.emit(CONNECTION_FAILED, new HASensors.FailureDescriptor(entity, description));
lastPublished = LastPublished.FAILED;
currentRecoveryStartTime = null;
} else {
long nextAttemptTime =
Math.max(
serviceFailedStabilizationDelay - currentFailurePeriod,
serviceFailedStabilizationDelay - sinceLastUpPeriod);
schedulePublish(nextAttemptTime);
}
}
}
}
public class CalculatedStatus {
public final boolean connected;
public CalculatedStatus() {
HostAndPort endpoint = getConfig(ENDPOINT);
connected = Networking.isReachable(endpoint);
}
public String getDescription() {
Long lastUpTime = connectionLastUp.get();
Long lastDownTime = connectionLastDown.get();
Duration serviceFailedStabilizationDelay = getConnectionFailedStabilizationDelay();
Duration serviceRecoveredStabilizationDelay = getConnectionRecoveredStabilizationDelay();
return String.format(
"endpoint=%s; connected=%s; timeNow=%s; lastUp=%s; lastDown=%s; lastPublished=%s; "
+ "currentFailurePeriod=%s; currentRecoveryPeriod=%s",
getConfig(ENDPOINT),
connected,
Time.makeDateString(System.currentTimeMillis()),
(lastUpTime != null ? Time.makeDateString(lastUpTime) : "<never>"),
(lastDownTime != null ? Time.makeDateString(lastDownTime) : "<never>"),
lastPublished,
(currentFailureStartTime != null ? getTimeStringSince(currentFailureStartTime) : "<none>")
+ " (stabilization "
+ makeTimeStringRounded(serviceFailedStabilizationDelay)
+ ")",
(currentRecoveryStartTime != null
? getTimeStringSince(currentRecoveryStartTime)
: "<none>")
+ " (stabilization "
+ makeTimeStringRounded(serviceRecoveredStabilizationDelay)
+ ")");
}
}
private long getTimeDiff(Long recent, Long previous) {
return (previous == null) ? recent : (recent - previous);
}
private String getTimeStringSince(Long time) {
return time == null ? null : Time.makeTimeStringRounded(System.currentTimeMillis() - time);
}
}
