ViewVC Help

View File | Revision Log | Show Annotations | Revision Graph | Root Listing

root/i-scream/projects/cms/source/server/uk/org/iscream/cms/server/client/monitors/Heartbeat__Monitor.java

Comparing projects/cms/source/server/uk/org/iscream/cms/server/client/monitors/Heartbeat__Monitor.java (file contents):
Revision 1.3 by tdb, Mon Mar 5 23:14:30 2001 UTC vs.
Revision 1.28 by tdb, Sun Aug 1 10:40:45 2004 UTC

#	Line 1 | Line 1
1	+	/*
2	+	* i-scream central monitoring system
3	+	* http://www.i-scream.org
4	+	* Copyright (C) 2000-2002 i-scream
5	+	*
6	+	* This program is free software; you can redistribute it and/or
7	+	* modify it under the terms of the GNU General Public License
8	+	* as published by the Free Software Foundation; either version 2
9	+	* of the License, or (at your option) any later version.
10	+	*
11	+	* This program is distributed in the hope that it will be useful,
12	+	* but WITHOUT ANY WARRANTY; without even the implied warranty of
13	+	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
14	+	* GNU General Public License for more details.
15	+	*
16	+	* You should have received a copy of the GNU General Public License
17	+	* along with this program; if not, write to the Free Software
18	+	* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
19	+	*/
20	+
21		//---PACKAGE DECLARATION---
22	<	package uk.ac.ukc.iscream.client.monitors;
22	>	package uk.org.iscream.cms.server.client.monitors;
23
24		//---IMPORTS---
25		import java.util.HashMap;
26		import java.util.Iterator;
27	<	import uk.ac.ukc.iscream.client.*;
28	<	import uk.ac.ukc.iscream.core.*;
29	<	import uk.ac.ukc.iscream.util.*;
30	<	import uk.ac.ukc.iscream.componentmanager.*;
27	>	import java.util.StringTokenizer;
28	>	import uk.org.iscream.cms.server.client.*;
29	>	import uk.org.iscream.cms.server.core.*;
30	>	import uk.org.iscream.cms.util.*;
31	>	import uk.org.iscream.cms.server.componentmanager.*;
32
33		/**
34	<	* This Monitor watches heartbeats
34	>	* This Monitor watches heartbeats.
35	>	* It generates an alert when a heartbeat that is expected
36	>	* does not arrive. Unlike all the other monitors, this one
37	>	* is driven by an event *not* occuring, rather than an
38	>	* event occuring. This means it must be actively checking
39	>	* for missing heartbeat's, and thus has an extra inner class
40	>	* thread.
41		*
42	+	* This originally took "heartbeat" packets, but they've now
43	+	* been deprecated. Instead we look at UDP packets, or, rather
44	+	* the lack of them :-)
45	+	*
46		* @author  $Author$
47		* @version $Id$
48		*/
49	<	public class Heartbeat__Monitor implements PluginMonitor, Runnable {
49	>	public class Heartbeat__Monitor extends MonitorSkeleton {
50
51		//---FINAL ATTRIBUTES---
52
#	Line 24 | Line 55 | public class Heartbeat__Monitor implements PluginMonit
55		*/
56		public final String REVISION = "$Revision$";
57
58	+	/**
59	+	* A description of this monitor
60	+	*/
61		public final String DESC = "Monitors Heartbeats.";
62
63	+	/**
64	+	* The default (used if not configured) period at
65	+	* which to check for old heartbeats. (in seconds)
66	+	*/
67		public final int DEFAULT_CHECK_PERIOD = 60;
68
69		//---STATIC METHODS---
70
71		//---CONSTRUCTORS---
72	<
72	>
73	>	/**
74	>	* Constructs a new Heartbeat monitor, and starts off
75	>	* the worker thread.
76	>	*/
77		public Heartbeat__Monitor() {
78	<	new Thread(this).start();
78	>	super();
79	>	createInitialHosts();
80	>	new HeartbeatWorker().start();
81		}
82
83		//---PUBLIC METHODS---
84
85	<	public void run() {
86	<	ConfigurationProxy cp = ConfigurationProxy.getInstance();
87	<	while(true) {
88	<	// this cycle period of this monitor's checks
89	<	int checkPeriod = 0;
90	<	try {
47	<	checkPeriod = Integer.parseInt(cp.getProperty(_name, "Monitor.Heartbeat.checkPeriod"));
48	<	} catch (PropertyNotFoundException e) {
49	<	checkPeriod = DEFAULT_CHECK_PERIOD;
50	<	_logger.write(toString(), Logger.WARNING, "Monitor.Heartbeat.checkPeriod value unavailable using default of " + checkPeriod + " seconds");
51	<	} catch (NumberFormatException e) {
52	<	checkPeriod = DEFAULT_CHECK_PERIOD;
53	<	_logger.write(toString(), Logger.WARNING, "Erronous Monitor.Heartbeat.checkPeriod value in configuration using default of " + checkPeriod + " seconds");
54	<	}
55	<
56	<	// perform the checks (use HB hash, although they *should* be the same)
57	<	Iterator i = _hostsHB.keySet().iterator();
58	<	while(i.hasNext()) {
59	<	// get host
60	<	String source = (String) i.next();
61	<	// check it
62	<	analyseHB(source);
63	<	}
64	<
65	<	// wait a while
66	<	try {Thread.sleep(checkPeriod * 1000);} catch (InterruptedException e) {}
67	<	}
68	<	}
69	<
70	<	// only use attribute num 0 :)
71	<	public void analyseHB(String source) {
72	<	ConfigurationProxy cp = ConfigurationProxy.getInstance();
73	<	Register reg = (Register) _hostsReg.get(source);
74	<
75	<	// get host's HB interval (seconds)
76	<	// this should always exist, thus we set to 0
77	<	int hostHBinterval = 0;
78	<	try {
79	<	hostHBinterval = Integer.parseInt(cp.getProperty("Host."+source, "TCPUpdateTime"));
80	<	} catch (PropertyNotFoundException e) {
81	<	hostHBinterval = 0;
82	<	_logger.write(toString(), Logger.WARNING, "TCPUpdateTime value unavailable using default of " + hostHBinterval + " seconds");
83	<	} catch (NumberFormatException e) {
84	<	hostHBinterval = 0;
85	<	_logger.write(toString(), Logger.WARNING, "Erronous TCPUpdateTime value in configuration using default of " + hostHBinterval + " seconds");
86	<	}
87	<
88	<	// get host's last HB time (seconds)
89	<	long lastHeartbeat = ((HeartbeatHolder) _hostsHB.get(source)).getLastHeartbeat();
90	<	// time since last heartbeat (seconds)
91	<	long timeSinceLastHB = (System.currentTimeMillis()/1000) - lastHeartbeat;
92	<
93	<	// find out the threshold level we're at
94	<	int result = checkAttributeThreshold(timeSinceLastHB, reg);
95	<
96	<	// decide what threshold level we're on, if we've changed, record that
97	<	if (result != reg.getLastThresholdLevel(0)) {
98	<	reg.setLastThresholdLevel(0, result);
99	<	}
100	<
101	<	// as long as this isn't a normal level
102	<	if(reg.getLastThresholdLevel(0) != Alert.thresholdNORMAL) {
103	<	// if the time since the last alert is more than the time for
104	<	// its timeout, fire an alert, escalate the alert
105	<	long timeout = reg.getLastAlertTimeout(0);
106	<	if ((timeout > 0) && (reg.getTimeLastSent(0) > 0)) {
107	<	if((System.currentTimeMillis() - reg.getTimeLastSent(0)) > timeout) {
108	<	int lastAlert = reg.getLastAlertLevel(0);
109	<	reg.escalateAlert(0);
110	<	reg.setTimeLastSent(0, System.currentTimeMillis());
111	<	reg.setLastAlertTimeout(0, reg.getAlertTimeout(reg.getLastAlertLevel(0), 0));
112	<	// -- SEND
113	<	fireAlert(source, timeSinceLastHB, reg, lastAlert);
114	<	}
115	<	// if we don't have a timeout configured...we got STRAIGHT to the next level
116	<	} else {
117	<	int lastAlert = reg.getLastAlertLevel(0);
118	<	reg.escalateAlert(0);
119	<	reg.setTimeLastSent(0, System.currentTimeMillis());
120	<	reg.setLastAlertTimeout(0, reg.getAlertTimeout(reg.getLastAlertLevel(0), 0));
121	<	// -- SEND
122	<	fireAlert(source, timeSinceLastHB, reg, lastAlert);
123	<	}
124	<
125	<	// we must be on ok, check the timeout value for this
126	<	} else {
127	<	// if we were on an OK alert before, then we don't do anything
128	<	// but if we weren't we only set OK, once the timeout of the last
129	<	// alert has occourd
130	<	if (reg.getLastAlertLevel(0) != Alert.alertOK) {
131	<	long timeout = reg.getLastAlertTimeout(0);
132	<	if ((timeout > 0) && (reg.getTimeLastSent(0) > 0)) {
133	<	if ((System.currentTimeMillis() - reg.getTimeLastSent(0)) > timeout) {
134	<	int lastAlert = reg.getLastAlertLevel(0);
135	<	reg.setLastAlertLevel(0, Alert.alertOK);
136	<	reg.setTimeLastSent(0, System.currentTimeMillis());
137	<	reg.setLastAlertTimeout(0, timeout);
138	<	// -- SEND
139	<	fireAlert(source, timeSinceLastHB, reg, lastAlert);
140	<	}
141	<	}
142	<	}
143	<	}
144	<	}
145	<
85	>	/**
86	>	* Analyse a packet of data. In this case, this will just
87	>	* register the fact that a heartbeat has arrived.
88	>	*
89	>	* @param packet The packet of data to analyse
90	>	*/
91		public void analysePacket(XMLPacket packet) {
92	<	if (packet.getParam("packet.attributes.type").equals("heartbeat")) {
93	<	String source = packet.getParam("packet.attributes.machine_name");
94	<	if (!_hostsHB.containsKey(source)) {
95	<	_hostsReg.put(source, new Register(source, _name, 1));
96	<	_hostsHB.put(source, new HeartbeatHolder());
92	>	String source = packet.getParam("packet.attributes.machine_name");
93	>	if(!checkBooleanConfig("Host." + source, "Monitor." + _name + ".enable")) {
94	>	return;
95	>	}
96	>	if (!_hosts.containsKey(source)) {
97	>	synchronized(this) {
98	>	_hosts.put(source, new HeartbeatHolder(new Register(source, _name)));
99		}
153	–	HeartbeatHolder lastHeartbeat = (HeartbeatHolder) _hostsHB.get(source);
154	–	lastHeartbeat.setLastHeartbeat(System.currentTimeMillis()/1000);
100		}
101	+	HeartbeatHolder lastHeartbeat = (HeartbeatHolder) _hosts.get(source);
102	+	lastHeartbeat.setLastHeartbeat(System.currentTimeMillis()/1000);
103		}
104
105		/**
106		* Overrides the {@link java.lang.Object#toString() Object.toString()}
107		* method to provide clean logging (every class should have this).
108		*
109	<	* This uses the uk.ac.ukc.iscream.util.NameFormat class
109	>	* This uses the uk.org.iscream.cms.util.NameFormat class
110		* to format the toString()
111		*
112		* @return the name of this class and its CVS revision
#	Line 180 | Line 127 | public class Heartbeat__Monitor implements PluginMonit
127
128		//---PRIVATE METHODS---
129
130	+	/**
131	+	* Checks whether the time since the last heartbeat
132	+	* is beyond the threshold(s).
133	+	*
134	+	* @param timeSinceLastHB a long time since the last heartbeat arrived
135	+	* @param reg the Register for this host
136	+	* @return the level which has been breached, if any
137	+	*/
138		private int checkAttributeThreshold(long timeSinceLastHB, Register reg) {
139		for(int thresholdLevel = Alert.thresholdLevels.length - 1; thresholdLevel >= 0; thresholdLevel--) {
140		if (reg.getThreshold(thresholdLevel) != -1.0) {
#	Line 188 | Line 143 | public class Heartbeat__Monitor implements PluginMonit
143		}
144		}
145		}
146	<	return 0;
146	>	return Alert.thresholdNORMAL;
147		}
148	+
149	+	/**
150	+	* Gets an initial list of hosts from the config
151	+	* and adds a fake set of heartbeats for them.
152	+	* If the hosts don't respond within the timeout
153	+	* period an alert will be raised.
154	+	*
155	+	* The effect of this is to allow us to know about
156	+	* hosts which weren't on when we started up, and
157	+	* will thus never have generated a heartbeat - yet
158	+	* will still want to know they're not responding.
159	+	*/
160	+	private void createInitialHosts() {
161	+	// get the initial list of hosts from the config
162	+	String initialHosts = "";
163	+	try {
164	+	initialHosts = _cp.getProperty(_name, "Monitor.Heartbeat.initialHosts");
165	+	} catch (PropertyNotFoundException e) {
166	+	// just leave initialHosts empty
167	+	_logger.write(Heartbeat__Monitor.this.toString(), Logger.DEBUG, "No initial list of hosts set, defaulting to none.");
168	+	}
169
170	<	private void fireAlert(String source, long timeSinceLastHB, Register reg, int lastAlert) {
171	<	int alertLevel = reg.getLastAlertLevel(0);
172	<	int thresholdLevel = reg.getLastThresholdLevel(0);
173	<	String currentValue = String.valueOf(timeSinceLastHB);
174	<	String attributeName = "Heartbeat";
175	<	String thresholdValue = String.valueOf(reg.getThreshold(thresholdLevel));
176	<	String time = Long.toString(reg.getAlertTimeout(reg.getLastAlertLevel(0), 0) / 1000);
177	<	if (thresholdLevel == Alert.thresholdNORMAL) {
178	<	thresholdValue = "-";
170	>	// parse through the initial hosts adding them
171	>	StringTokenizer st = new StringTokenizer(initialHosts, ";");
172	>	while (st.hasMoreTokens()) {
173	>	String source = st.nextToken();
174	>	// check if they already exist, don't want to add them twice
175	>	if (!_hosts.containsKey(source)) {
176	>	synchronized(this) {
177	>	_hosts.put(source, new HeartbeatHolder(new Register(source, _name)));
178	>	}
179	>	}
180	>	HeartbeatHolder lastHeartbeat = (HeartbeatHolder) _hosts.get(source);
181	>	// set a "fake" heartbeat
182	>	lastHeartbeat.setLastHeartbeat(System.currentTimeMillis()/1000);
183		}
204	–	if (alertLevel == Alert.alertOK) {
205	–	time = "0";
206	–	}
207	–	Alert alert = new Alert(alertLevel, lastAlert, thresholdLevel, source, thresholdValue, currentValue, attributeName, time);
208	–	_alerterQueue.add(alert);
209	–	_logger.write(toString(), Logger.DEBUG, "Fired alert for source:" + source + " at alert level:" + Alert.alertLevels[alertLevel] + " on:" + attributeName + " for threshold level:" + Alert.thresholdLevels[thresholdLevel] + " at:" +  currentValue + " exceeding threshold of:" +thresholdValue + " next alert sent in:" + time + "secs");
184		}
185
186		//---ACCESSOR/MUTATOR METHODS---
187	<
187	>
188	>	/**
189	>	* Returns a reference to the Queue we're getting data
190	>	* from. This is specific to this monitor.
191	>	*
192	>	* @return a reference to a Queue to get data from
193	>	*/
194	>	protected Queue getQueue() {
195	>	return MonitorManager.getInstance().getDataQueue();
196	>	}
197	>
198		//---ATTRIBUTES---
199
200		/**
#	Line 223 | Line 207 | public class Heartbeat__Monitor implements PluginMonit
207		* be changed to null for utility classes.
208		*/
209		private String _name = "Heartbeat";
210	<
210	>
211		/**
212	<	* This holds a reference to the
229	<	* system logger that is being used.
212	>	* A reference to the configuration proxy in use
213		*/
214	<	private Logger _logger = ReferenceManager.getInstance().getLogger();
214	>	private ConfigurationProxy _cp = ConfigurationProxy.getInstance();
215
216	<	private Queue _alerterQueue = ClientMain._alerterQueue;
216	>	/**
217	>	* A HashMap of hosts, with associated HeartbeatHolder's.
218	>	*/
219	>	private HashMap _hosts = new HashMap();
220
221		/**
222	<	* A reference to the configuration proxy in use
222	>	* A reference to the system logger.
223		*/
224	<	private ConfigurationProxy _cp = ConfigurationProxy.getInstance();
224	>	private Logger _logger = ReferenceManager.getInstance().getLogger();
225
240	–	private HashMap _hostsHB = new HashMap();
241	–	private HashMap _hostsReg = new HashMap();
242	–
226		//---STATIC ATTRIBUTES---
227
228		//---INNER CLASSES---
229	<
229	>
230	>	/**
231	>	* This inner class simply holding some information
232	>	* about a specific host.
233	>	*/
234		private class HeartbeatHolder {
235
236	+	/**
237	+	* Construct a new HeartbeatHolder.
238	+	*/
239	+	public HeartbeatHolder(Register register) {
240	+	_register = register;
241	+	}
242	+
243	+	/**
244	+	* Set the time of the last heartbeat
245	+	*/
246		public void setLastHeartbeat(long lastHeartbeat) {
247		_lastHeartbeat = lastHeartbeat;
248		}
249
250	+	/**
251	+	* Get the time of the last heartbeat
252	+	*/
253		public long getLastHeartbeat() {
254		return _lastHeartbeat;
255		}
256
257	+	/**
258	+	* Get the Register
259	+	*/
260	+	public Register getRegister() {
261	+	return _register;
262	+	}
263	+
264	+	/**
265	+	* last heartbeat time
266	+	*/
267		private long _lastHeartbeat;
268	<	}
269	<
268	>
269	>	/**
270	>	* register ref
271	>	*/
272	>	private Register _register;
273	>	}
274	>
275	>	/**
276	>	* This worker thread just checks all the hosts and then
277	>	* waits a period of time before doing it again. It sends
278	>	* Alerts as required.
279	>	*/
280	>	private class HeartbeatWorker extends Thread {
281	>
282	>	/**
283	>	* The main run method of this worker thread. It simply
284	>	* checks through all the hosts it has stored, running
285	>	* the analyseHB method on each. It then removes any
286	>	* that have passed a FINAL, and waits a (configured)
287	>	* length of time before doing it again.
288	>	*/
289	>	public void run() {
290	>	ConfigurationProxy cp = ConfigurationProxy.getInstance();
291	>	while(true) {
292	>	// this cycle period of this monitor's checks
293	>	int checkPeriod = 0;
294	>	try {
295	>	checkPeriod = Integer.parseInt(cp.getProperty(_name, "Monitor.Heartbeat.checkPeriod"));
296	>	} catch (PropertyNotFoundException e) {
297	>	checkPeriod = DEFAULT_CHECK_PERIOD;
298	>	_logger.write(Heartbeat__Monitor.this.toString(), Logger.WARNING, "Monitor.Heartbeat.checkPeriod value unavailable using default of " + checkPeriod + " seconds");
299	>	} catch (NumberFormatException e) {
300	>	checkPeriod = DEFAULT_CHECK_PERIOD;
301	>	_logger.write(Heartbeat__Monitor.this.toString(), Logger.WARNING, "Erronous Monitor.Heartbeat.checkPeriod value in configuration using default of " + checkPeriod + " seconds");
302	>	}
303	>
304	>	synchronized(Heartbeat__Monitor.this) {
305	>	// perform the checks (use HB hash, although they *should* be the same)
306	>	Iterator i = _hosts.keySet().iterator();
307	>	while(i.hasNext()) {
308	>	// get host
309	>	String source = (String) i.next();
310	>	// check it
311	>	boolean remove = analyseHB(source);
312	>	// remove it if it's passed a FINAL
313	>	if(remove) {
314	>	i.remove();
315	>	}
316	>	}
317	>	}
318	>
319	>	// wait a while
320	>	try {Thread.sleep(checkPeriod * 1000);} catch (InterruptedException e) {}
321	>	}
322	>	}
323	>
324	>	/**
325	>	* Analyses a given host's state, and if need be generates
326	>	* a relevant Alert. Note that it also checks if the last
327	>	* alert sent is FINAL, in which case it returns true to
328	>	* indicate removal of this host.
329	>	*
330	>	* @param source the host to check
331	>	* @return whether this host can be deleted
332	>	*/
333	>	private boolean analyseHB(String source) {
334	>	ConfigurationProxy cp = ConfigurationProxy.getInstance();
335	>	HeartbeatHolder hbHolder = (HeartbeatHolder) _hosts.get(source);
336	>	Register reg = hbHolder.getRegister();
337	>
338	>	// get host's HB interval (seconds)
339	>	// this should always exist, thus we set to 0
340	>	int hostHBinterval = 0;
341	>	try {
342	>	hostHBinterval = Integer.parseInt(cp.getProperty("Host."+source, "Host.UDPUpdateTime"));
343	>	} catch (PropertyNotFoundException e) {
344	>	hostHBinterval = 0;
345	>	_logger.write(Heartbeat__Monitor.this.toString(), Logger.WARNING, "UDPUpdateTime value unavailable using default of " + hostHBinterval + " seconds");
346	>	} catch (NumberFormatException e) {
347	>	hostHBinterval = 0;
348	>	_logger.write(Heartbeat__Monitor.this.toString(), Logger.WARNING, "Erronous UDPUpdateTime value in configuration using default of " + hostHBinterval + " seconds");
349	>	}
350	>
351	>	// get host's last HB time (seconds)
352	>	long lastHeartbeat = hbHolder.getLastHeartbeat();
353	>	// time since last heartbeat (seconds)
354	>	long timeSinceLastHB = (System.currentTimeMillis()/1000) - lastHeartbeat;
355	>	// time since (or until if negative) the expected heartbeat
356	>	long timeSinceExpectedHB = timeSinceLastHB - (long) hostHBinterval;
357	>
358	>	// best do a check in case the expected heartbeat is in the future
359	>	if(timeSinceExpectedHB < 0) {
360	>	timeSinceExpectedHB = 0;
361	>	}
362	>
363	>	// find out the threshold level we're at
364	>	int newThreshold = checkAttributeThreshold(timeSinceExpectedHB, reg);
365	>
366	>	// process the alert
367	>	Heartbeat__Monitor.this.processAlert(newThreshold, "Heartbeat", reg, source, String.valueOf(timeSinceExpectedHB));
368	>
369	>	if(reg.getLastAlertLevel() == Alert.alertFINAL) {
370	>	return true;
371	>	}
372	>	return false;
373	>	}
374	>	}
375		}

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines