RemoraJ: Extensible Java Bytecode Agent for Optimizing Java App Performance
RemoraJ is an extensible java profiling agent which uses bytecode instrumentation to intercept java IPC calls with minimal overhead. RemoraJ monitors java apps by tracking IPC calls such as: HTTP, WebServices, JMS, JDBC, Sockets, WebSockets, Kafka, I/O Streams and more. The goal is to capture inter-JVM calls, I/O, messages, exceptions, timings which are valuable during troubleshooting and performance optimization of java apps.
RemoraJ consists of two main runtime components:
Remora Java Agent
: loaded into JVM, app server, instruments bytecode, emits traces and metrics via pluggable advicesRemora Streams Agent
: forwards the traces generated by the java agent to one of the supported analytics platforms below
Platforms for Analyzing RemoraJ Traces
- Nastel XRay -- SaaS Platform for AIOps & transaction tracking
- Integrate traces into ELK, Splunk or other platform via agent output class implementation
Both agents communicate via a memory mapped queue stored on a filesystem see tnt-data-source.xml
for details. The Streams Agent sends traces using JSON over HTTPS.
NOTE: Streams Agent is based on tnt4j-streams project designed to process, parse and stream time series data over a number of transports such as: HTTPS, Kafka, JMS, MQTT, MQ, files, etc.
RemoraJ Intercepts (Advices)
RemoraJ supports the following intercepts (advices):
"ApacheHttpClientAdvice",
"ApacheLegacyHttpClientAdvice",
"BankBenchmarkAdvice",
"RemoraControlAdvice",
"EjbRemoteAdvice",
"HttpUrlConnectionAdvice",
"JavaxServletAdvice",
"JBossAdvice",
"JBossServletAdvice",
"JDBCConnectionAdvice",
"JDBCStatementAdvice",
"JDBCCallableStatementAdvice",
"JMSCreateConnectionAdvice",
"JMSReceiveAdvice",
"JMSSendAdvice",
"KafkaConsumerAdvice",
"KafkaProducerAdvice",
"KafkaConsumerClientAdvice",
"SimpleTest",
"SimpleTestConstructor",
"SpringServiceAdvice",
"SpringExceptionAdvice",
"SpringTransactionAdvice",
"WebLogicAdvice",
"WebsocketSendAdvice",
"WebsocketEndpointAdvice",
"WebsocketReceiveAdvice",
"WebsocketSessionAdvice",
"IBMAdapterRSA",
"WASAdvice"
Installing RemoraJ
You must configure the Remora Java agent and Streams agent to run RemoraJ. Your should have a working data repository and a streaming access token. Below is the outline of RemoraJ setup:
- Configure the Remora Java Agent for your JVM (captures java call traces, metrics)
- Start your application or app server (with remora agent)
Please complete the following steps if you decide to send traces to Nastel XRay:
- Obtain your account with Nastel XRay. Make sure to get your streaming access token
- Configure the Streams Agent (you will need your streaming access token here)
- Start the Streams Agent (forwards traces to your data repository associated with your streaming access token)
- Run your Java apps with RemoraJ Agent enabled
- Login to your Nastel XRay Dashboard dashboard to analyze your traces
Prerequisites
Java 8+ runtime (IBM WebSphere 8.5.5+, Tomcat 8+, jBoss 7.2+ should be already running Java 8. Make sure JVM is Java 1.8 or above).
The following IBM WebSphere runtime jar files are required to build remora-websphere
advices:
com.ibm.jaxws.thinclient_8.5.0.jar
com.ibm.ws.runtime.jar
com.ibm.ws.webcontainer.jar
rsahelpers.jar
com.ibm.ws.admin.core.jar
Place these jar files in remora-websphere/lib
folder before running a build.
Remora Java Agent
Using -javaagent option
IBM WebSphere
-
Option 1: using IBM WebSphere console.
- Step 1: Navigate to
Application servers > [Your server name] > Process definition > Java Virtual Machine
- Step 2: Edit field "Generic JVM arguments"
- Step 3: Add
-javaagent:[<install_dir>/remora-<version>]/remora.jar
- Step 4: Restart IBM WebSphere
- Step 5: Run and configure Streams forwarding agent
- Step 1: Navigate to
-
Option 2: editing
server.xml
properties manually- Step 1: Navigate to
<USER_DIR>/IBM/WebSphere/<Server>/profiles/<App Server>/config/cells/<Cell>/nodes/<Node>/servers/<Server>/server.xml
- Step 2: Edit node
/process:Server/processDefinitions/jvmEntries
parameter@genericJvmArguments
- Step 3: Edit the path to where your
remora.jar
situated
<jvmEntries xmi:id="JavaVirtualMachine_1183122130078" verboseModeClass="false" verboseModeGarbageCollection="false" verboseModeJNI="false" initialHeapSize="512" maximumHeapSize="2056" runHProf="false" hprofArguments="" genericJvmArguments="-javaagent:<install_dir>/remora-<version>/remora.jar" executableJarFileName="" disableJIT="false">
- Step 1: Navigate to
IBM WAS Liberty
- Step 1: Edit or create
jvm.options
file in the folderwlp/usr/servers/<serverName>/
- Step 2: Add lines:
-javaagent:<install_dir>/remora-<version>/remora.jar -Dremora.path
- Step 3: Edit the path to where your
remora.jar
situated - Step 4: Edit or create
bootstrap.properties
in the folderwlp/usr/servers/<serverName>/
- Step 5: Add line:
org.osgi.framework.bootdelegation=com.jkoolcloud.remora.*
JBoss Application Server
Standalone mode
- Step 1: Edit
bin/standalone.bat[.sh]
- Step 2: Add line:
winx: set "JAVA_OPTS=%JAVA_OPTS% -javaagent:<install_dir>/remora-<version>/remora.jar"
unix: JAVA_OPTS="$JAVA_OPTS -javaagent:<install_dir>/remora-<version>/remora.jar"
- Step 3: Edit the path to where your
remora.jar
is located
Domain mode
- Step 1: Edit
domain/configuration/host.xml
- Step 2: Edit tag
<servers><jvm>
- Step 3: Add
<option value="-javaagent:<install_dir>/remora-<version>/remora.jar"/>
<jvm name="default">
<jvm-options>
</jvm-options>
<option value="-agentlib:jdwp=transport=dt_socket,address=5007,server=y,suspend=n"/>
<option value="-javaagent:<install_dir>/remora-<version>/remora.jar"/>
</jvm>
- Step 4: Edit the path to where your
remora.jar
is located
Standalone Java Application
To run you standalone application with RemoraJ add option -javaagent:<install_dir>/remora-<version>/remora.jar
to your run script or command line:
java -javaagent:<install_dir>/remora-<version>/remora.jar -jar <jar-file-name>.jar
Configure Streams Agent (Optional)
As for version 0.1.9 Streams will be configured automatically on first run. For more sophisticated control you can follow these instructions:
- Step 1: (Optional) Go to
<install_dir>/remora-<version>/tnt4j-streams/config
- Step 2: (Optional) Edit
tnt4j-streams.properties
and setup your access token (event.sink.factory.EventSinkFactory.prod.Token
) - Step 3: (Optional) More settings in
<install_dir>/remora-<version>/tnt4j-streams/remora-streamer/tnt-data-source.xml
- Step 4: (Optional) Edit
<property name="FileName" value="../../queue"/>
to point to your RemoraJ queue directory.
NOTE: RemoraJ queue directory hosts files containing java traces produced by remora java agent. The traces are read by the Streams Agent and forwarded to your data repository.
Remora Agent Configuration
RemoraJ configuration file is located config
folder, file named remora.properties
.
See relevant comments in the file for advanced config.
Agent VM Id
To identify your VM: set system property (java -Dremora.vmid=MyVmName
) to any readable token identifying the process remora agent is
attached to. remora.vmid
is set to ManagementFactory.getRuntimeMXBean().getName()
by default.
NOTE: remora.vmid
value might be not available on every VM's implementation and may be platform dependent.
Application Id
To identify your application: set system property (java -Dremora.appl.name=MyApp1
). All emmited traces are tagged with the specified app name. Application name is automatically set based on deployed app name when remora agent is running in the application server context such as jBoss.
NOTE: Default remora.appl.name
is set to java
if not explicitly specified.
Agent Trace Output
By default, RemoraJ java agent writes its traces to a high-performance memmory mapped persistent store (file) backed by com.jkoolcloud.remora.core.output.ChronicleOutput
. This store is used as a communication channel between RemoraJ java agent and the Streams Agent.
There are several trace outputs available below:
com.jkoolcloud.remora.core.output.NullOutput
-- no output (null output)com.jkoolcloud.remora.core.output.SysOutOutput
--System.out
(console)com.jkoolcloud.remora.core.output.ChronicleOutput
-- persistent memory mapped file-based output (default)
File system queue is created only with com.jkoolcloud.remora.core.output.ChronicleOutput
. Set system property (java -Dremora.output
) with the trace output class reference (e.g. -Dremora.output=com.jkoolcloud.remora.core.output.SysOutOutput
).
Advice Trace Filters
Advices can have trace filters. Filters determine which trace are include vs. excluded. You configure the filter by defining one in remora.properties
.
<prefix for filter definition - filter>.<unique filter name>.<filter property>
for example:
filter.myDefinedFilter.type=com.jkoolcloud.remora.filters.ClassFilter
filter.myDefinedFilter.mode=INCLUDE/EXCLUDE
filter.myDefinedFilter.classes=java.net.SocketInputStream
to apply the filter(s) to advice, add as advice
property:
com.jkoolcloud.remora.advices.<advice name>.filters=<unique filter name>;<unique filter name2>;<unique filter name3>
Running Streams Agent
Once the Remora Java Agent is running and the Streams Agent configured you can run <install_dir>/remora-<version>/tnt4j-streams/remora-streamer/run.sh(run.bat)
to start the Streams Agent and begin forwarding java traces to your trace output. Running first time you will be asked for a repository token. This tokes is latter written in file "token", if you'd like to change.
Troubleshooting and Logging
If there is some problems running you can always check the logs. Logging might cause seriuos overhead, so by default the logging option is turned off.
In order to turn on the logging you need to change configuration file, or you can turn it on remotelly if you are using remora-control
module.
To turn on logging you need to change option com.jkoolcloud.remora.advices.BaseTransformers.logging
to true
in the file remora.properties
. You can also turn on logging for individual advices.
Each advice creates it own logging file. You can find all files in your remora's folder under log
-- <install_dir>/remora-<version>/log
.
Supported Analytics Platforms
- Nastel XRay -- AIOps & transaction tracking. Supports SaaS/On-prem/Cloud
- Integrate traces into ELK, Splunk or other platform via agent output class implementation