Stats and their classes
Generally everything can be monitored with MoSKito. However, some things need to be monitored more often than others. And some things needs to be monitored so often, that they come out of the box. Here a list of Stats objects that come out of the box. Stat objects represents stats (statistics) about something that can be count. A single stat object represents usage of a single monitored object, that can be an url, a method, a business case, a thread, a memory pool, a disk or file, a cpu or anything else worth monitoring.
Below you will see general stat class and what is monitored.
The following diagram gives a short overview of stat classes that come out of the box.
How to read the table:
| Value | Meaning | Shortcuts | Unit | Notes |
|---|---|---|---|---|
| name of the value of the appropriate class | What does this attribute actually measure | What shortcuts can be used for generic access to this value | What is the unit (time, amount etc) | Something special to think about? |
Please note, even the Value is the name of the StatValue Object used inside the Stats object*, the accessor methods can differ. For example, to add time to total time in totalRequests you have to call addExecutionTime. Also an accessor can do different type of things, for example:
public void addRequest() {
totalRequests.increase();
currentRequests.increase();
maxCurrentRequests.setValueIfGreaterThanCurrentAsLong(currentRequests.getValueAsLong());
}
public void addExecutionTime(long time) {
totalTime.increaseByLong(time);
lastRequest.setValueAsLong(time);
minTime.setValueIfLesserThanCurrentAsLong(time);
maxTime.setValueIfGreaterThanCurrentAsLong(time);
}
Please refer to the documentation and examples on how to use the stat objects correctly (or use built-in annotations)
A huge bunch of stats is request oriented (method calls, urls, web-actions etc). The class net.anotheria.moskito.core.predefined.RequestOrientedStats provides general functionality for the request oriented stats. The concrete subclasses add only specific handling.
| Value | Meaning | Shortcuts | Unit | Notes |
|---|---|---|---|---|
| totalRequests | Number of times this request has been executed | TR, REQ | Amount | |
| totalTime | Total time spent in this request | TT, TIME | Time, NS | |
| currentRequests | Number of requests currently in this request | CR | Amount | |
| maxCurrentRequests | Maximal recorded number of executions of this request concurrently. | MCR | Amount | |
| errors | Number of recorded errors. | ERR | Amount | |
| lastRequest | Duration of last request. | LAST | Time, NS | |
| minTime | Minimal duration of this request. | MIN | Time, NS | |
| maxTime | Maximal duration of this request. | MAX | Time, NS | |
| averageTime | Average duration of this request. | AVG | Time, NS | Calculated by totalTime/totalRequests |
Used for service like components (Service, API etc), also for Dynamic Proxies (See net.anotheria.moskito.core.dynamic.MoskitoInvokationProxy) and @MonitorClass, @Monitor CDI and AOP annotations.
ActionStats are used by/for action driven frameworks such as struts or ano-maf, but similar to ServiceStats, differs from the RequestStats just by name. The point of having different names(classes) is that it allows the stats to be presented in different decorators and not getting mixed in webui.
ServletStats is the extension of the request oriented stats for http servlets and have three additional values. Each method (get,put,head...) will get its own stats object with all the values. See net.anotheria.moskito.web.MoskitoHttpServlet for examples or easy to extend counting servlet.
| Value | Meaning | Shortcuts | Unit | Notes |
|---|---|---|---|---|
| ioExceptions | Number of io exceptions in this method. | IOExc | Amount | |
| servletExceptions | Number of servlet exceptions in this method. | SEExc | Amount | |
| runtimeExceptions | Number of runtime exceptions in this method. | RTExc | Amount |
Same as Servlet Stats, but for Filters. See net.anotheria.moskito.web.MoskitoFilter, net.anotheria.moskito.web.filters.DomainFilter, net.anotheria.moskito.web.filters.RefererFilter or net.anotheria.moskito.web.filters.RequestURIFilter for usage example or building your own.
Environment and JVM Stats are a huge group of stats that are not produced by the Application directly, but by its environment, either the JVM or the operating system or hardware. Most of those stats are gathered from appropriate MBeans.
Memory is one of the most important parameters of any system in production.
Based on Runtime.freeMemory(), maxMemory() and totalMemory() this simple object gives some overview about the memory. MemoryStats are builtin and available ootb. See net.anotheria.moskito.core.util.BuiltInMemoryProducer.
A MemoryStats object exists for each stat: free, max and total. The MemoryStats are updated once a minute by a Timer.
| Value | Meaning | Shortcuts | Unit | Notes |
|---|---|---|---|---|
| current | Current amount of memory | CUR | Size, Bytes | |
| min | Min amount of memory. | MIN | Size, Bytes | |
| max | Max amount of memory. | MAX | Size, Bytes |
A small example how it looks like in the MoSKito Inspect:
This stats object is based on the MemoryPoolMXBean and provides information about available memory pools. MemoryPoolStats are builtin and available ootb. See net.anotheria.moskito.core.util.BuiltInMemoryPoolProducer. MemoryPoolStats are updated once a minute by a Timer.
MemoryPoolStats are much more precise than the MemoryStats. See java.lang.management.MemoryUsage for details.
| Value | Meaning | Shortcuts | Unit | Notes |
|---|---|---|---|---|
| init | Initial amount | INIT | Size, Bytes | MemoryUsage.init |
| used | Used in current interval. | USED | Size, Bytes | MemoryUsage.used |
| minUsed | Min used in current interval | MIN_USED | Size, Bytes | |
| maxUsed | Max used in current interval | MAX_USED | Size, Bytes | |
| commited | Commited in current interval. | COMMITED | Size, Bytes | MemoryUsage.commited |
| minCommited | Min commited. | MIN_COMMITED | Size, Bytes | |
| maxCommited | Max commited. | MAX_COMMITED | Size, Bytes | |
| max | Max available. | MAX | Size, Bytes | MemoryUsage.max |
Sometimes its hard to calculate the different heap pools to one heap usage. VirtualMemoryPoolStats do exactly that by combining different pools by their heap-ness into heap and non-heap pools. VirtualMemoryPoolStats are builtin and available ootb. See net.anotheria.moskito.core.util.BuiltInMemoryPoolVirtualProducer. Virtual Memory Pool has exactly the same values (cumulated) as underlying pools and therefore the same values and shortcuts.
Threads are no less important than memory. The data is based on java.lang.management.ThreadMXBean.
This stat object is for simple thread counting. It always has one stat. The data is updated once a minute. See net.anotheria.moskito.core.util.BuiltInThreadCountProducer.
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| started | Threads started in the interval. | STARTED | Amount | |
| current | Currently running threads. | CUR, CURRENT | Amount | |
| daemon | Currently running daemon threads. | DAEMON | Amount | |
| minCurrent | Minimum amount of currently running threads. | MIN, MINCURRENT, MIN CUR | Amount | |
| maxCurrent | Maximum amount of currently running threads. | MAX, MAXCURRENT, MAX CUR | Amount |
ThreadStateStats are ment to hold amount of threads in each state (BLOCKED, WAITING etc). However, the corresponding producer is currently disabled as of 2.0.1. If it would be reenabled, it would deliver following data for each possible STATE:
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| current | Currently running threads in this state. | CUR | Amount | |
| min | Minimum amount of currently running threads in this state. | MIN | Amount | |
| max | Maximum amount of currently running threads in this state. | MAX | Amount |
OSStats are based on com.sun.management.UnixOperatingSystemMXBean and therefore only supported on Unix Platforms.
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| openFiles | Currently open files in the process. | openfiles, open files | Amount | |
| maxOpenFiles | Max open files by the process. | minopenfiles, min open files | Amount | |
| minOpenFiles | Min open files by the process. | maxopenfiles, max open files | Amount | |
| maxSupportedOpenFiles | Max supported open files. | maxsupportedopenfiles | Amount | This value doesn't (shouldn't) change after start. |
| processCpuTime | CPU Time consumed by this process | cputime, cpu time | Time, Millis | |
| freePhysicalMemory | Free physical memory. | free memory, free | Size, Bytes | |
| totalPhysicalMemory | Total physical memory. | total memory, total | Size, Bytes | This value doesn't (shouldn't) change after start. |
| processors | Number of processors | Amount | This value doesn't (shouldn't) change after start. |
RuntimeStats are useful if you want to know how long your application is running without logging into the machine. They are not that relevant for monitoring. They are based on java.lang.management.RuntimeMXBean.
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| processName | Currently running threads in this state. | process, name, processname | String | |
| startTime | Minimum amount of currently running threads in this state. | starttime | Time, Millis since 01.01.1970 | |
| uptime | Maximum amount of currently running threads in this state. | uptime | Time, Millis since start. |
Keeping refers to everything that is stored temporary somewhere (probably in memory) and is proceeded somehow. Keeping related stats can be used for capacity planing and monitoring and for throughput monitoring. Following Stats belong into this category:
CacheStats are good for monitoring ... caches! Alas if you want a collection of good and easy to use monitored caches - check out ano-prise-caches.
Not all cache implementations will provide every possible stat value.
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| requests | Get request to the cache. | REQ | Amount | |
| hits | Hit requests. | HIT | Amount | |
| writes | Number of writes of new objects to the cache. | WR | Amount | |
| garbageCollected | Number of items that were garbage collected. | GC | Amount | Only supported by SoftReferenceCaches or similar. |
| rolloverCount | Number of items that were rolled over. | RO | Amount | Only supported by fixed size caches with rollover. (RoundRobin) |
| expiredCount | Number of items that were expired. | EX | Amount | Only supported by expiry caches. |
| filteredCount | Number of items that were in cache but not returned due to filtering | FI | Amount | Only supported by caches that support filtering. |
| cacheFullCount | Number of times that the cache were full and refused write. | FU | Amount | Only supported by not rolling over caches. |
| deletes | Number of items removed from the cache. | DEL | Amount | |
| hitrate | Percentage of hits among gets (hits/requests) | HR | Calculated |
A queue in moskitos sense is something between two components, where one component (feeder) add stuff from one end, and the other (processor) works on the other end and removes it. Of course it can be many2many.
Despite their beautifulness, QueueStats are somewhat out of date and are subject to be reworked in near future.
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| requests | Request to put something in the queue. | Amount | ||
| enqueued | Successful request to put something in the queue. | Amount | ||
| dequeued | Successfully processed and therefore removed items. | Amount | ||
| empty | Number of times the processor found the queue empty and had nothing to do. | Amount | ||
| totalSize | Max possible number of elements in the queue. | Amount | ||
| lastSize | Size at the moment of the last operation (which means number of elements in the queue) | Amount | ||
| sumOfSizes | Sum of element count in the queue at updates (puts). | Amount | ||
| maxSize | Max number of elements in the queue. | Amount | ||
| minSize | Min number of elements in the queue. | Amount |
Used for monitoring of QueuingSystems. As for now, only one such system is known: net.anotheria.anoprise.processor.QueuedMultiProcessor.
Despite their beautifulness, QueuingSystemStats are somewhat out of date and are subject to be reworked in near future.
| Value | Meaning | Shortcut | Units | Note |
|---|---|---|---|---|
| serversSize | Amount of servers processing queue. | Amount | ||
| queueSize | Number of elements in the queue. | Amount | ||
| arrived | Number of arrived elements. | Amount | ||
| serviced | Number of serviced elements. | Amount | ||
| errors | Number of errors. | Amount | ||
| waited | Number of elements waiting for processing. | Amount | ||
| throwedAway | Number of elements thrown away due to overflow. | Amount | ||
| waitingTime | Waiting time of the elements. | Time, Millis | ||
| waitingTimeMin | Min waiting time of an element.. | Time, Millis | ||
| waitingTimeMax | Max waiting time of an element. | Time, Millis | ||
| servicingTime | Time spent servicing the elements. | Time, Millis | ||
| servicingTimeMin | Min time was spent to serve an element. | Time, Millis | ||
| servicingTimeMax | Max time that was spent to serve an element. | Time, Millis |
Storages are basically wrapper to maps. And storage stats monitor accesses to it.
| Value | Meaning | Shortcuts | Units | Note |
|---|---|---|---|---|
| gets | calls to get() method. | G | Amount | |
| missedGets | gets that returned null | mG | Amount | |
| puts | calls to put() method | P | Amount | |
| overwritePuts | calls to put() methods that overwrote an existing value | oP | Amount | |
| removes | calls to remove() method | RM | Amount | |
| noopRemoves | calls to remove() method that had no effect (no element there) | noRM | Amount | |
| size | size of the storage | SZ | Amount | |
| containsKeyCalls | Calls to containsKey method | CKC | Amount | |
| containsKeyHits | Calls to containsKey method that returned not null | CKH | Amount | Reason: Calls to containsKey method are expensive, its good to monitor them and their success. |
| containsValueCalls | Calls to containsValue method | CVC | Amount | |
| containsValueHits | Calls to containsValue method that returned not null | CVH | Amount | Reason: Calls to containsValue method are VERY expensive, its good to monitor them and their success. |
| missedGets Ratio | missedGets/gets | mG R | Ratio, 0..1 | Calculated |
| hitGets Ratio | (gets-missedGets)/gets | hG R | Ratio, 0..1 | Calculated |
| overwritingPut Ratio | overwritePuts/puts | oP R | Ratio, 0..1 | Calculated |
| newPuts Ratio | (puts-overwritePuts)/puts | nP R | Ratio, 0..1 | Calculated |
| noopRemoves Ratio | noopRemoves/removes | noRM R | Ratio, 0..1 | Calculated |
| PutGet Ratio | puts/gets | PG R | Ratio, | Calculated, if >1 you are putting more elements than you read, this can be unhealthy. |
| PutRemove Ratio | puts/removes | PRM R | Ratio, | Calculated, if >1 you are putting more elements than you remove, this can be a memory leak. |
| containsKeyHitRate | containsKeyHits/containsKeyCalls | CK HR | Ratio, 0..1 | Calculated |
| containsKeyValueRate | containsValueHits/containsValueCalls | CV HR | Ratio, 0..1 | Calculated |