Our Garbage collection log analysis REST API provides powerful micrometrics. Using these micrometrics, you can predict and forecast the application’s availability and performance characteristics.

Watch this short video clip to see how and where our API can be used.

How to invoke Garbage Collection log analysis API?

Invoking Garbage Collection log analysis is very simple:

  1. Register with us. We will email you the API key. This is a one-time setup process. Note: If you have purchased enterprise version with API, you don’t have to register. API key will be provided to you as part of installation instruction.
  2. Post HTTP request to https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}
  3. The body of the HTTP request should contain the Garbage collection log that needs to be analyzed.
  4. HTTP Response will be sent back in JSON format. JSON has several important stats about the GC log. Primary element to look in the JSON response is: “isProblem“. This element will have value to be “true” if any memory/performance problems has been discovered. “problem” element will contain the detailed description of the memory problem.

CURL command

Assuming your Java GC log file is located in “./my-app-gc.log,” then CURL command to invoke the API is:

curl -X POST --data-binary @./my-app-gc.log https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY} --header "Content-Type:text"

It can’t get any more simpler than that? Isn’t it? 🙂

Compression

GC log files are quite large in size. For fast and efficient processing, we recommend you to compress and send the GC log files. When you are compressing the GC log, you need to pass ‘Content-Encoding’ element in the HTTP Header element or in the URL parameter.

Say suppose you are compressing GC log file into ‘zip’ format, then you can invoke the API with HTTP header element

curl -X POST --data-binary @./my-app-gc.zip "https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}" --header "Content-Type:text" --header "Content-Encoding:zip"

or you can also invoke the API with ‘Content-Encoding’ element in the URL parameter

curl -X POST --data-binary @./my-app-gc.zip "https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}&Content-Encoding=zip" --header "Content-Type:text"

We support following compression formats :

zip, gz, xz, z, lzma, deflate, sz, lz4, zstd, bz2, tar

You may use the one of your choice. Whatever compression format you used for compressing the GC log files should be passed in ‘Content-Encoding’ element.

Note:  use the option “–data-binary” in the CURL instead of using “–data” option. In “–data” new line breaks will be not preserved in the request. New Line breaks should be preserved for legitimate parsing.

Downloading GC Log from remote location

Say suppose, you have stored your Java GC log files in a remote location like AWS S3 bucket. In those cases, if you want to analyze your GC logs then you will have to download files to your local machine and then invoke the API with GC log file. To make it simple and automation easier we have introduced ‘location’ parameter in our API. You can pass http(s) URL location to the ‘location’ parameter and invoke the API. When you do it, API will download file from the remote location, analyze the data and return back the response. For more details refer to this article.

curl -X POST "https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}&location={GC-LOG-HTTP-URL}" --header "Content-Type:text"

Other Tools

You can also invoke the API using any webservice client tools such as: SOAP UI, Postman Browser Plugin,…..

postman

Fig: POSTing GC logs through PostMan plugin

Sample Response

{
{
  "isProblem": true,
  "problem": [
    "Our analysis tells that Full GCs are consecutively running in your application. It might cause intermittent OutOfMemoryErrors or degradation in response time or high CPU consumption or even make application unresponsive.",
    "Our analysis tells that your application is suffering from long running GC events. 4 GC events took more than 10 seconds. Long running GCs are unfavourable for application's performance.",
    "342 times application threads were stopped for more than 10 seconds."
  ],
  "jvmHeapSize": {
    "youngGen": {
      "allocatedSize": "7.5 gb",
      "peakSize": "6 gb"
    },
    "oldGen": {
      "allocatedSize": "22.5 gb",
      "peakSize": "22.5 gb"
    },
    "metaSpace": {
      "allocatedSize": "1.04 gb",
      "peakSize": "48.52 mb"
    },
    "total": {
      "allocatedSize": "30 gb",
      "peakSize": "28.5 gb"
    }
  },
  "gcStatistics": {
    "totalCreatedBytes": "249.49 gb",
    "measurementDuration": "7 hrs 32 min 52 sec",
    "startTimestamp": "2018-12-15T15:58:37.622+0100",
    "endTimestamp": "2018-12-15T15:58:44.622+0100",
    "avgAllocationRate": "9.4 mb/sec",
    "avgPromotionRate": "1.35 mb/sec",
    "minorGCCount": "62",
    "minorGCTotalTime": "1 min 19 sec",
    "minorGCAvgTime": "1 sec 274 ms",
    "minorGCAvgTimeStdDeviation": "2 sec 374 ms",
    "minorGCMinTIme": "0",
    "minorGCMaxTime": "13 sec 780 ms",
    "minorGCIntervalAvgTime": "7 min 25 sec 442 ms",
    "fullGCCount": "166",
    "fullGCTotalTime": "14 min 11 sec 620 ms",
    "fullGCAvgTime": "5 sec 130 ms",
    "fullGCAvgTimeStdDeviation": "5 sec 207 ms",
    "fullGCMinTIme": "120 ms",
    "fullGCMaxTime": "57 sec 880 ms",
    "fullGCIntervalAvgTime": "2 min 19 sec 104 ms"
  },
  "gcKPI": {
    "throughputPercentage": 99.952,
    "averagePauseTime": 750.232,
    "maxPauseTime": 57880
 },
  "gcDurationSummary": {
    "groups": [
      {
        "start": "0",
        "end": "6",
        "numberOfGCs": 212
      },
      {
        "start": "6",
        "end": "12",
        "numberOfGCs": 4
      },
      {
        "start": "12",
        "end": "18",
        "numberOfGCs": 2
      },
      {
        "start": "42",
        "end": "48",
        "numberOfGCs": 1
      },
      {
        "start": "54",
        "end": "60",
        "numberOfGCs": 1
      }
    ]
  },
  "gcCauses": [
    {
      "cause": "Allocation Failure",
      "count": 57
    },
    {
      "cause": "Concurrent Mode Failure",
      "count": 162
    },
    {
      "cause": "Full GC - Allocation Failure",
      "count": 369
    }
  ],
  "commandLineFlags": " -XX:CMSInitiatingOccupancyFraction=65 -XX:+CMSScavengeBeforeRemark -XX:CMSWaitDuration=2000 -XX:ConcGCThreads=8 -XX:+DisableExplicitGC -XX:GCLogFileSize=104857600 -XX:+HeapDumpOnOutOfMemoryError -XX:InitialHeapSize=32212254720 -XX:InitialTenuringThreshold=4 -XX:+ManagementServer -XX:MaxHeapSize=32212254720 -XX:MaxTenuringThreshold=4 -XX:NewRatio=3 -XX:NumberOfGCLogFiles=10 -XX:OldPLABSize=16 -XX:ParGCCardsPerStrideChunk=32768 -XX:+PrintClassHistogram -XX:+PrintGC -XX:+PrintGCApplicationStoppedTime -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+PrintPromotionFailure -XX:+PrintTenuringDistribution -XX:SurvivorRatio=3 -XX:ThreadStackSize=334 -XX:+UnlockDiagnosticVMOptions -XX:+UseCMSInitiatingOccupancyOnly -XX:+UseCompressedClassPointers -XX:+UseCompressedOops -XX:+UseConcMarkSweepGC -XX:+UseGCLogFileRotation -XX:+UseParNewGC ",
  "heapTuningTips": [
    "It looks like you have over allocated Metaspace size. During entire run, Metaspace's peak utilization was only 4.53% of the allocated size. You can consider lowering the Metaspace Size."
  ],
  "tipsToReduceGCTime": [
    {
      "issue": "15.34% of GC time (i.e 54 min 13 sec 710 ms) is caused by 'Concurrent Mode Failure'. The CMS collector uses one or more garbage collector threads that run simultaneously with the application threads with the goal of completing the collection of the tenured generation before it becomes full. In normal operation, the CMS collector does most of its tracing and sweeping work with the application threads still running, so only brief pauses are seen by the application threads. However, if the CMS collector is unable to finish reclaiming the unreachable objects before the tenured generation fills up, or if an allocation cannot be satisfied with the available free space blocks in the tenured generation, then the application is paused and the collection is completed with all the application threads stopped. The inability to complete a collection concurrently is referred to as concurrent mode failure and indicates the need to adjust the CMS collector parameters. Concurrent mode failure typically triggers Full GC..",
      "solution": "The concurrent mode failure can either be avoided by increasing the tenured generation size or initiating the CMS collection at a lesser heap occupancy by setting CMSInitiatingOccupancyFraction to a lower value and setting UseCMSInitiatingOccupancyOnly to true. CMSInitiatingOccupancyFraction should be chosen carefuly, setting it to a low value will result in too frequent CMS collections."
    }
  ],
  "throughputPercentage": 99.996,
  "responseId": "8296b5c3-25c7-4157-92df-a54d9083bab7",
  "graphs": {
        "heapAfterGCGraph": "https://graphs.gceasy.io/archived/2019/04/14/--heap_usage_after_gc-aad50e08-951f-4eca-9cc2-e2b296ba3f8b.png",
        "heapBeforeGCGraph": "https://graphs.gceasy.io/archived/2019/04/14/--heap_usage_before_gc-868d5a56-c8b6-4125-bf8d-dc557937033a.png",
        "gcDurationGraph": "https://graphs.gceasy.io/archived/2019/04/14/--gc_duration-1048abf9-8833-44f1-a990-9a6dd24e9a94.png",
        "gcPauseDurationGCGraph": "https://graphs.gceasy.io/archived/2019/04/14/--gc_pause_duration-ef8b210e-1d08-4b5f-be28-e2d54a43732d.png"
 }
  "webReport": "http://gceasy.io/my-gc-report.jsp?p=YXJjaGl2ZWQvMjAxOS8wNC8xNC8tLWFwaS1lMDk0YTM0ZS1jM2ViLTRjOWEtODI1NC1mMGRkMTA3MjQ1Y2M1YjBjMTMzNy0yMDQyLTQ4NzAtYmI5My01NmEwZjUzOTg3M2YudHh0LS0=&channel=API",
  "pdfReport": "http://gceasy.io/my-gc-report.jsp?p=YXJjaGl2ZWQvMjAxOS8wNC8xNC8tLWFwaS1lMDk0YTM0ZS1jM2ViLTRjOWEtODI1NC1mMGRkMTA3MjQ1Y2M1YjBjMTMzNy0yMDQyLTQ4NzAtYmI5My01NmEwZjUzOTg3M2YudHh0LS0=&channel=API&isPrinterFriendly=true"
}

JSON Response Elements

Element Description
responseId Unique transaction Id that is generated for every response. This is used for any debugging or diagnosis purposes
webReport Hyperlink to view the visual web report of this GC analysis on web browser.
pdfReport Hyperlink to download the PDF report of this GC analysis from your web browser.
isProblem true’ is returned if any memory problems are found. ‘false’ is returned if no memory problems are found. This element can be used to build any alerts for proactive GC monitoring
problem Description of the memory problem is reported in this element. Like what type of memory problem, it is. What are the side-effects and symptoms it might cause
jvmHeapSize The data points in this section is gathered from the GC log, thus It may or may not match with the size that is specified by the JVM system properties (i.e. –Xmx, -Xms,…). Say you have configured total heap size (i.e. –Xmx) as 2gb, whereas at runtime if JVM has allocated only 1gb, then in this report you are going to see the allocated size as 1gb only.
youngGen
> allocatedSize Young Generation’s allocated size (i.e. specified at the JVM level)
> peakSize Young Generation’s peak utilization size at runtime
oldGen
> allocatedSize Old Generation’s allocated size (i.e. specified at the JVM level)
> peakSize Old Generation’s peak utilization size at runtime
metaSpace
> allocatedSize Metaspace’s allocated size (i.e. specified at the JVM level)
> peakSize MetaSpace’s peak utilization size at runtime
permGen
> allocatedSize Perm Generation’s allocated size (i.e. specified at the JVM level)
> peakSize Perm Generation’s peak utilization size at runtime
total
> allocatedSize Total allocated heap size (i.e. specified at the JVM level) includes Young + Old + Perm (or Metaspace)
> peakSize Peak utilization of the heap size at runtime
gcStatistics GC statistics summary is based on ‘real’ time reported in the GC Logs. For more details on difference between user, sys, real time refer to http://blog.gceasy.io/2016/04/06/gc-logging-user-sys-real-which-time-to-use/
> totalCreatedBytes Total amount of objects created by the application
> measurementDuration Total time duration application has been running
> startTimestamp Timestamp at which first GC event was reported
> endTimestamp Timestamp in which last GC event was reported
> avgAllocationRate Objects creation rate by the application. When more objects are allocated, young generation gets filled up quickly, and Minor GC runs more frequently.
> avgPromotionRate Objects promoted rate from Young Generation to Old Generation. When more objects are promoted to Old generation, Full GC will be run more frequently.
> minorGCCount Number of Minor GCs occurred in the application
> minorGCTotalTime Total time spent in Minor GCs
> minorGCAvgTime Average time taken by the Minor GCs (i.e. total time taken by all minor GCs / number of minor GCs)
> minorGCAvgTimeStdDeviation Standard Deviation of the Minor GC average times
> minorGCMinTIme Minimum Time of all the minor GCs
> minorGCMaxTime Maximum time of all the minor GCs
> minorGCIntervalAvgTime Average time interval between minor GCs
> fullGCCount Number of Full GCs occurred in the application
> fullGCTotalTime Total time spent in Full GCs
> fullGCAvgTime Average time taken by the Full GCs (i.e. total time taken by all full GCs / number of full GCs)
> fullGCAvgTimeStdDeviation Standard Deviation of the Full GC average times
> fullGCMinTime Minimum Time of all the full GCs
> fullGCMaxTime Maximum time of all the full GCs
> fullGCIntervalAvgTime Average time interval between Full GCs
gcKPI Key Performance Indicators of this GC analysis
>throughputPercentage Percentage of time spent in processing real transactions vs time spent in GC activity. Higher percentage is a good indication that GC overhead is low. One should aim for high throughput.
>averagePauseTime Average GC Pause time in milliseconds
>maxPauseTime Maximum GC Pause time in milliseconds
gcDurationSummary
> groups Array of GC Duration summary group. Example of a group:
{
“start”: “0.2”,
“end”: “0.3”,
“numberOfGCs”: 5
}, Indicates that 5 GCs completed between 0.2 second and 0.3 seconds
>> start Group’s start time reported in seconds
>> end Group’s end time reported in seconds
>> numberOfGCs Number of GCs that completed between ‘start’ time and ‘end’ time
graphs Graph images for this GC log analysis can be found in the child elements
> heapAfterGCGraph contains URL to the Heap usage after GC graph image
> heapBeforeGCGraph contains URL to the Heap usage before GC graph image
> gcDurationGraph contains URL to the GC Duration graph image
> gcPauseDurationGCGraph contains URL GC Pause Duration graph image
gcCauses array  reporting the GC Causes
> cause The reason which triggered the GC. Values could be: Concurrent Mode Failure, Allocation Failure, System.gc() calls
> count Number of time GC occurred because of this reason
commandLineFlags JVM arguments that were passed to the application
heapTuningTips Array. Optional Element. Provides tips to tune your JVM heap size
tipsToReduceGCTime Array. Optional Element. Provides tips to reduce GC time
 >issue Indicates the reason what caused the GC
 >solution Solution to reduce this GC cause

Tagging the report

Suppose you are analyzing GC logs across hundreds of JVM. You want to tag each report with the hostname, date,… so that it will be handy to know which report belongs to which host. This can be achieved by passing request parameter ‘fileName‘ in the GC API endpoint i.e.

https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}&fileName={YOUR_TAG}.

Response JSON contains ‘webReport’ element. This element contains an URL. When you paste this URL in the browser, you will see the visual report of the GC log that was parsed through API. Now on the right top corner of the report in the browser, you will be able to see the tag you had passed in the API.

Say suppose if you pass fileName=hostABC8080 in the API endpoint, this is how it will show up in the report.

tag

Fig: GCeasy report with the tag passed in API

XML Response

By default API response in sent in JSON format. However if you would like to get response in XML format, you can pass ‘accept’ HTTP header element with value ‘xml’

Curl –X POST --data-binary @./my-app-gc.log https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY} –-header "accept:xml"

or you can also invoke the API with ‘accept’ element in the URL parameter

Curl –X POST --data-binary @./my-app-gc.log https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}&accept=xml

Graph Images

Few customers would like to see GCeasy API response contains URLs to graph images. Customers uses these graph images for following use cases:

a. Building dashboards of GC log analysis.
b. Integrating GC analysis in to CI/CD pipeline.
c. Automatic performance reports generation across hundreds of nodes.

Inorder to enable graphs in API response, you need to pass ‘graphs=true’ in the API request header, as show below:

Curl –X POST --data-binary @./my-app-gc.log https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}&graphs=true

Then you will start to see API response contain URLs to graphs, as shown below:

 "graphs": {
        "heapAfterGCGraph": "https://graphs.gceasy.io/archived/2019/04/14/--heap_usage_after_gc-aad50e08-951f-4eca-9cc2-e2b296ba3f8b.png",
        "heapBeforeGCGraph": "https://graphs.gceasy.io/archived/2019/04/14/--heap_usage_before_gc-868d5a56-c8b6-4125-bf8d-dc557937033a.png",
        "gcDurationGraph": "https://graphs.gceasy.io/archived/2019/04/14/--gc_duration-1048abf9-8833-44f1-a990-9a6dd24e9a94.png",
        "gcPauseDurationGCGraph": "https://graphs.gceasy.io/archived/2019/04/14/--gc_pause_duration-ef8b210e-1d08-4b5f-be28-e2d54a43732d.png"
 }

Normalized Units

Response elements which has memory sizes are sent back in kb, mb, gb, tb units. Example:

“allocatedSize”: “2.9 gb”,
“peakSize”: “1.36 gb”

Similarly elements with time are sent back in hours, minutes, seconds, milliseconds units. Example:

“minorGCTotalTime”: “1 min 9 sec 408 ms”,
“minorGCMaxTime”: “1 sec 65 ms”,

However if you would like to get the memory units to be normalized in ‘bytes’ and time in ‘milliseconds’, you can pass ‘normalizeUnits’ URL parameter with value to be ‘true’

Curl –X POST --data-binary @./my-app-gc.log https://api.gceasy.io/analyzeGC?apiKey={YOUR_API_KEY}&normalizeUnits=true

When you invoke API with option, response elements will be normalized as shown below:

“allocatedSize”: “3.113877504E9”,
“peakSize”: “2.95285248E9”

“minorGCTotalTime”: “69408.36”,
“minorGCMaxTime”: “1065.17”,

Time based Filtering

Sometimes users you want to analyze GC logs for a certain time interval instead of analyzing the entire Java GC Log. Here are the use cases for it:

1. If GC logs are not configured to roll over (which is typically the case in most organizations), one single GC log file will end up having the information of all GC events. Certain applications would have been started days/weeks back. Thus, one single GC log file would end up having all those days/weeks’ worth of GC events. The user might want to analyze only for last 24 hours’ worth of GC activities

2. Each application will have high traffic volume and low traffic volume time in a day. User might want to analyze the GC only during peak volume time periods. Say, if the user is from the financial industry, and he wants to monitor GC activities only during COW (Critical Online Window) i.e. 9:00am to 6:00pm.

To do such time based processing, refer to this article.

Relative to Absolute Timestamp

In your GC log file, timestamps might be printed in relative format i.e.

[0.209s]   .....

But however if you would like the GCeasy graphs to be painted in absolute timestamp, you might consider passing the ‘initialTimestamp‘ parameter in the API request. More details on this feature can be found here.

Key elements in API Response

GCeasy’s JSON APIs are used for application monitoring, analyzing code quality during CI/CD pipeline and several other purposes. API response contains rich set of information (i.e. lot of elements). In this article, we have highlighted few key elements in the API response. If values of these elements exceed or drops below certain threshold, you might consider raising alerts/warnings or breaking the build.

Best Practice: Building alerts

For production monitoring or for performance tests, you can build alerts around the JSON response.

a. If “isProblem” element has value to be “true” then alert can generated with the content from “problem” element. 

b. You can also setup thresholds on the elements like: “fullGCCount”, “fullGCTotalTime”, “fullGCAvgTime”, “fullGCAvgTimeStdDeviation”, “fullGCMinTIme”, “fullGCMaxTime”,
“fullGCIntervalAvgTime”. If any these elements value exceeds your threshold then alerts can be triggered.