double Checkable::CalculateExecutionTime(const CheckResult::Ptr& cr)
{
if (!cr)
return 0;
return cr->GetExecutionEnd() - cr->GetExecutionStart();
}
void Checkable::ProcessCheckResult(const CheckResult::Ptr& cr, const MessageOrigin::Ptr& origin)
{
{
ObjectLock olock(this);
m_CheckRunning = false;
}
double now = Utility::GetTime();
if (cr->GetScheduleStart() == 0)
cr->SetScheduleStart(now);
if (cr->GetScheduleEnd() == 0)
cr->SetScheduleEnd(now);
if (cr->GetExecutionStart() == 0)
cr->SetExecutionStart(now);
if (cr->GetExecutionEnd() == 0)
cr->SetExecutionEnd(now);
if (!origin || origin->IsLocal()) {
Log(LogDebug, "Checkable")
<< "No origin or local origin for object '" << GetName()
<< "', setting " << IcingaApplication::GetInstance()->GetNodeName()
<< " as check_source.";
cr->SetCheckSource(IcingaApplication::GetInstance()->GetNodeName());
}
Endpoint::Ptr command_endpoint = GetCommandEndpoint();
/* override check source if command_endpoint was defined */
if (command_endpoint && !GetExtension("agent_check")) {
Log(LogDebug, "Checkable")
<< "command_endpoint found for object '" << GetName()
<< "', setting " << command_endpoint->GetName()
<< " as check_source.";
cr->SetCheckSource(command_endpoint->GetName());
}
/* agent checks go through the api */
if (command_endpoint && GetExtension("agent_check")) {
ApiListener::Ptr listener = ApiListener::GetInstance();
if (listener) {
/* send message back to its origin */
Dictionary::Ptr message = ClusterEvents::MakeCheckResultMessage(this, cr);
listener->SyncSendMessage(command_endpoint, message);
}
return;
}
bool reachable = IsReachable();
bool notification_reachable = IsReachable(DependencyNotification);
ASSERT(!OwnsLock());
ObjectLock olock(this);
CheckResult::Ptr old_cr = GetLastCheckResult();
ServiceState old_state = GetStateRaw();
StateType old_stateType = GetStateType();
long old_attempt = GetCheckAttempt();
bool recovery = false;
if (old_cr && cr->GetExecutionStart() < old_cr->GetExecutionStart())
return;
/* The ExecuteCheck function already sets the old state, but we need to do it again
* in case this was a passive check result. */
SetLastStateRaw(old_state);
SetLastStateType(old_stateType);
SetLastReachable(reachable);
long attempt = 1;
std::set<Checkable::Ptr> children = GetChildren();
if (!old_cr) {
SetStateType(StateTypeHard);
} else if (cr->GetState() == ServiceOK) {
if (old_state == ServiceOK && old_stateType == StateTypeSoft) {
SetStateType(StateTypeHard); // SOFT OK -> HARD OK
recovery = true;
}
if (old_state != ServiceOK)
recovery = true; // NOT OK -> SOFT/HARD OK
ResetNotificationNumbers();
SetLastStateOK(Utility::GetTime());
/* update reachability for child objects in OK state */
if (!children.empty())
OnReachabilityChanged(this, cr, children, origin);
} else {
if (old_attempt >= GetMaxCheckAttempts()) {
SetStateType(StateTypeHard);
} else if (old_stateType == StateTypeSoft && old_state != ServiceOK) {
SetStateType(StateTypeSoft);
attempt = old_attempt + 1; //NOT-OK -> NOT-OK counter
} else if (old_state == ServiceOK) {
SetStateType(StateTypeSoft);
attempt = 1; //OK -> NOT-OK transition, reset the counter
} else {
attempt = old_attempt;
}
switch (cr->GetState()) {
case ServiceOK:
/* Nothing to do here. */
break;
case ServiceWarning:
SetLastStateWarning(Utility::GetTime());
break;
case ServiceCritical:
SetLastStateCritical(Utility::GetTime());
break;
case ServiceUnknown:
SetLastStateUnknown(Utility::GetTime());
break;
}
/* update reachability for child objects in NOT-OK state */
if (!children.empty())
OnReachabilityChanged(this, cr, children, origin);
}
if (!reachable)
SetLastStateUnreachable(Utility::GetTime());
SetCheckAttempt(attempt);
ServiceState new_state = cr->GetState();
SetStateRaw(new_state);
bool stateChange = (old_state != new_state);
if (stateChange) {
SetLastStateChange(now);
/* remove acknowledgements */
if (GetAcknowledgement() == AcknowledgementNormal ||
(GetAcknowledgement() == AcknowledgementSticky && new_state == ServiceOK)) {
ClearAcknowledgement();
}
/* reschedule direct parents */
BOOST_FOREACH(const Checkable::Ptr& parent, GetParents()) {
if (parent.get() == this)
continue;
ObjectLock olock(parent);
parent->SetNextCheck(Utility::GetTime());
}
}
bool remove_acknowledgement_comments = false;
if (GetAcknowledgement() == AcknowledgementNone)
remove_acknowledgement_comments = true;
bool hardChange = (GetStateType() == StateTypeHard && old_stateType == StateTypeSoft);
if (stateChange && old_stateType == StateTypeHard && GetStateType() == StateTypeHard)
hardChange = true;
bool is_volatile = GetVolatile();
if (hardChange || is_volatile) {
SetLastHardStateRaw(new_state);
SetLastHardStateChange(now);
}
if (new_state != ServiceOK)
TriggerDowntimes();
Host::Ptr host;
Service::Ptr service;
tie(host, service) = GetHostService(this);
CheckableType checkable_type = CheckableHost;
if (service)
checkable_type = CheckableService;
/* statistics for external tools */
Checkable::UpdateStatistics(cr, checkable_type);
bool in_downtime = IsInDowntime();
bool send_notification = hardChange && notification_reachable && !in_downtime && !IsAcknowledged();
if (!old_cr)
send_notification = false; /* Don't send notifications for the initial state change */
if (old_state == ServiceOK && old_stateType == StateTypeSoft)
send_notification = false; /* Don't send notifications for SOFT-OK -> HARD-OK. */
if (is_volatile && old_state == ServiceOK && new_state == ServiceOK)
send_notification = false; /* Don't send notifications for volatile OK -> OK changes. */
bool send_downtime_notification = (GetLastInDowntime() != in_downtime);
SetLastInDowntime(in_downtime);
olock.Unlock();
if (remove_acknowledgement_comments)
RemoveCommentsByType(CommentAcknowledgement);
Dictionary::Ptr vars_after = new Dictionary();
vars_after->Set("state", new_state);
vars_after->Set("state_type", GetStateType());
vars_after->Set("attempt", GetCheckAttempt());
vars_after->Set("reachable", reachable);
if (old_cr)
cr->SetVarsBefore(old_cr->GetVarsAfter());
cr->SetVarsAfter(vars_after);
olock.Lock();
SetLastCheckResult(cr);
bool was_flapping, is_flapping;
was_flapping = IsFlapping();
if (GetStateType() == StateTypeHard)
UpdateFlappingStatus(stateChange);
is_flapping = IsFlapping();
olock.Unlock();
// Log(LogDebug, "Checkable")
// << "Flapping: Checkable " << GetName()
// << " was: " << (was_flapping)
// << " is: " << is_flapping)
// << " threshold: " << GetFlappingThreshold()
// << "% current: " + GetFlappingCurrent()) << "%.";
OnNewCheckResult(this, cr, origin);
/* signal status updates to for example db_ido */
OnStateChanged(this);
String old_state_str = (service ? Service::StateToString(old_state) : Host::StateToString(Host::CalculateState(old_state)));
String new_state_str = (service ? Service::StateToString(new_state) : Host::StateToString(Host::CalculateState(new_state)));
if (hardChange || is_volatile) {
OnStateChange(this, cr, StateTypeHard, origin);
Log(LogNotice, "Checkable")
<< "State Change: Checkable " << GetName() << " hard state change from " << old_state_str << " to " << new_state_str << " detected." << (is_volatile ? " Checkable is volatile." : "");
} else if (stateChange) {
OnStateChange(this, cr, StateTypeSoft, origin);
Log(LogNotice, "Checkable")
<< "State Change: Checkable " << GetName() << " soft state change from " << old_state_str << " to " << new_state_str << " detected.";
}
if (GetStateType() == StateTypeSoft || hardChange || recovery || is_volatile)
ExecuteEventHandler();
if (send_downtime_notification)
OnNotificationsRequested(this, in_downtime ? NotificationDowntimeStart : NotificationDowntimeEnd, cr, "", "");
if (!was_flapping && is_flapping) {
OnNotificationsRequested(this, NotificationFlappingStart, cr, "", "");
Log(LogNotice, "Checkable")
<< "Flapping: Checkable " << GetName() << " started flapping (" << GetFlappingThreshold() << "% < " << GetFlappingCurrent() << "%).";
NotifyFlapping(origin);
} else if (was_flapping && !is_flapping) {
OnNotificationsRequested(this, NotificationFlappingEnd, cr, "", "");
Log(LogNotice, "Checkable")
<< "Flapping: Checkable " << GetName() << " stopped flapping (" << GetFlappingThreshold() << "% >= " << GetFlappingCurrent() << "%).";
NotifyFlapping(origin);
} else if (send_notification)
OnNotificationsRequested(this, recovery ? NotificationRecovery : NotificationProblem, cr, "", "");
}
void CheckResultReader::ProcessCheckResultFile(const String& path) const
{
CONTEXT("Processing check result file '" + path + "'");
String crfile = String(path.Begin(), path.End() - 3); /* Remove the ".ok" extension. */
std::ifstream fp;
fp.exceptions(std::ifstream::badbit);
fp.open(crfile.CStr());
std::map<String, String> attrs;
while (fp.good()) {
std::string line;
std::getline(fp, line);
if (line.empty() || line[0] == '#')
continue; /* Ignore comments and empty lines. */
size_t pos = line.find_first_of('=');
if (pos == std::string::npos)
continue; /* Ignore invalid lines. */
String key = line.substr(0, pos);
String value = line.substr(pos + 1);
attrs[key] = value;
}
/* Remove the checkresult files. */
if (unlink(path.CStr()) < 0)
BOOST_THROW_EXCEPTION(posix_error()
<< boost::errinfo_api_function("unlink")
<< boost::errinfo_errno(errno)
<< boost::errinfo_file_name(path));
if (unlink(crfile.CStr()) < 0)
BOOST_THROW_EXCEPTION(posix_error()
<< boost::errinfo_api_function("unlink")
<< boost::errinfo_errno(errno)
<< boost::errinfo_file_name(crfile));
Checkable::Ptr checkable;
Host::Ptr host = Host::GetByName(attrs["host_name"]);
if (!host) {
Log(LogWarning, "CheckResultReader")
<< "Ignoring checkresult file for host '" << attrs["host_name"] << "': Host does not exist.";
return;
}
if (attrs.find("service_description") != attrs.end()) {
Service::Ptr service = host->GetServiceByShortName(attrs["service_description"]);
if (!service) {
Log(LogWarning, "CheckResultReader")
<< "Ignoring checkresult file for host '" << attrs["host_name"]
<< "', service '" << attrs["service_description"] << "': Service does not exist.";
return;
}
checkable = service;
} else
checkable = host;
CheckResult::Ptr result = new CheckResult();
String output = CompatUtility::UnEscapeString(attrs["output"]);
std::pair<String, Value> co = PluginUtility::ParseCheckOutput(output);
result->SetOutput(co.first);
result->SetPerformanceData(PluginUtility::SplitPerfdata(co.second));
result->SetState(PluginUtility::ExitStatusToState(Convert::ToLong(attrs["return_code"])));
if (attrs.find("start_time") != attrs.end())
result->SetExecutionStart(Convert::ToDouble(attrs["start_time"]));
else
result->SetExecutionStart(Utility::GetTime());
if (attrs.find("finish_time") != attrs.end())
result->SetExecutionEnd(Convert::ToDouble(attrs["finish_time"]));
else
result->SetExecutionEnd(result->GetExecutionStart());
checkable->ProcessCheckResult(result);
Log(LogDebug, "CheckResultReader")
<< "Processed checkresult file for object '" << checkable->GetName() << "'";
/* Reschedule the next check. The side effect of this is that for as long
* as we receive check result files for a host/service we won't execute any
* active checks. */
checkable->SetNextCheck(Utility::GetTime() + checkable->GetCheckInterval());
}
void ElasticsearchWriter::AddCheckResult(const Dictionary::Ptr& fields, const Checkable::Ptr& checkable, const CheckResult::Ptr& cr)
{
String prefix = "check_result.";
fields->Set(prefix + "output", cr->GetOutput());
fields->Set(prefix + "check_source", cr->GetCheckSource());
fields->Set(prefix + "exit_status", cr->GetExitStatus());
fields->Set(prefix + "command", cr->GetCommand());
fields->Set(prefix + "state", cr->GetState());
fields->Set(prefix + "vars_before", cr->GetVarsBefore());
fields->Set(prefix + "vars_after", cr->GetVarsAfter());
fields->Set(prefix + "execution_start", FormatTimestamp(cr->GetExecutionStart()));
fields->Set(prefix + "execution_end", FormatTimestamp(cr->GetExecutionEnd()));
fields->Set(prefix + "schedule_start", FormatTimestamp(cr->GetScheduleStart()));
fields->Set(prefix + "schedule_end", FormatTimestamp(cr->GetScheduleEnd()));
/* Add extra calculated field. */
fields->Set(prefix + "latency", cr->CalculateLatency());
fields->Set(prefix + "execution_time", cr->CalculateExecutionTime());
if (!GetEnableSendPerfdata())
return;
Array::Ptr perfdata = cr->GetPerformanceData();
CheckCommand::Ptr checkCommand = checkable->GetCheckCommand();
if (perfdata) {
ObjectLock olock(perfdata);
for (const Value& val : perfdata) {
PerfdataValue::Ptr pdv;
if (val.IsObjectType<PerfdataValue>())
pdv = val;
else {
try {
pdv = PerfdataValue::Parse(val);
} catch (const std::exception&) {
Log(LogWarning, "ElasticsearchWriter")
<< "Ignoring invalid perfdata for checkable '"
<< checkable->GetName() << "' and command '"
<< checkCommand->GetName() << "' with value: " << val;
continue;
}
}
String escapedKey = pdv->GetLabel();
boost::replace_all(escapedKey, " ", "_");
boost::replace_all(escapedKey, ".", "_");
boost::replace_all(escapedKey, "\\", "_");
boost::algorithm::replace_all(escapedKey, "::", ".");
String perfdataPrefix = prefix + "perfdata." + escapedKey;
fields->Set(perfdataPrefix + ".value", pdv->GetValue());
if (pdv->GetMin())
fields->Set(perfdataPrefix + ".min", pdv->GetMin());
if (pdv->GetMax())
fields->Set(perfdataPrefix + ".max", pdv->GetMax());
if (pdv->GetWarn())
fields->Set(perfdataPrefix + ".warn", pdv->GetWarn());
if (pdv->GetCrit())
fields->Set(perfdataPrefix + ".crit", pdv->GetCrit());
if (!pdv->GetUnit().IsEmpty())
fields->Set(perfdataPrefix + ".unit", pdv->GetUnit());
}
}
}
