AEM CQ5 Introduction: TAR Compaction for AEM Performance optimization

TARMK Compaction

We require this when we see growth in TarMk files in the repository. As we all know that tar file data is never overwritten and they always do append the data. So the disk usage increases even when only updating existing data. To avoid such increase in repository,

AEM provides this Tar Compaction. This mechanism will reclaim disk space by removing obsolete data from the repository.

We have two kinds of Tar Compactions.

➢ Online Compaction (AEM 6.2 Doesn’t recommend it anymore)

➢ Offline Compaction

Even before jumping on to these details, let us discussion a bit on automatic compaction triggered manually.

Revision Clean Up:

The automatic compaction can be triggered manually in the Operations Dashboard via a maintenance job called Revision Clean Up.

To Start Revision Clean Up you need to:

❏ Go to AEM Welcome Screen.rm

❏ In the main AEM window, go to Tools → Operations → Dashboard → Maintenance

❏ Or directly browse below.

http://localhost:4502/libs/granite/operations/content/maintenance.html

❏ You see the screen like below.

❏ Now click on Daily Maintenance Window

❏ Hover over the Revision Clean Up window and press the start button like below.

❏ Click on the “Run” icon as seen on the above screenshot.

❏ The icon will turn orange to indicate Revision Clean Up job is running.

You can stop it at any time by hovering the mouse over the icon and pressing the Stop button.

Invoking Revision Garbage Collection via the JMX Console

➔ Open the JMX console as

http://localhost:4502/system/console/jmx

➔ Click RevisionGarbageCollection MBean

➔ In the next window, click startRevisionGC() and then Invoke to start the Revision Garbage Collection job.

TAR OFFLINE COMPACTION:

NOTE: Never do offline compaction when the repository is up and running. Not even checkpoints should be done while the server is up and running.

The procedure is called offline compaction because the repository needs to be shut down in order to properly run the Oak-run tool.

For faster compaction of the Tar files and situations where normal garbage collection does not work, Adobe provides a manual Tar compaction tool called Oak-run. It can be downloaded at the following location: http://mvnrepository.com/artifact/org.apache.jackrabbit/oak-run/

Note: We use Oak-run jar depending upon the version of our current Oak in the AEM server. So make sure we have the right version of oak-run jar.

The procedure to run the tool is:

➔ Always make sure we have a recent working backup of the AEM instance

➔ Shutdown AEM

➔ Use the tool to find old checkpoints:

java -jar oak-run.jar checkpoints <aem-inst-folder>/crx-quickstart/repository/segmentstore

➔ When you run the above command you should be able to see the references of the nodes like below which we need to clear them before compaction takes place.

➔ Now delete the unreferenced checkpoints using below command.

java -jar oak-run.jar checkpoints <aem-inst-folder>/crx-quickstart/repository/segmentstore rm-unreferenced

➔ Finally run the compaction and wait for it to complete

java -jar oak-run.jar compact <aem-inst-folder>/crx-quickstart/repository/segmentstore

➔ You should be see something like below when you run the above command.

➔ It should display all files under segmentstore directory. After a while you should be able to see Cleaning up message like below.

➔ Once it cleans up you should be able to see the below message.

➔ It will display the tarfiles again with less count after compaction.

We can create a log file to track all these changes in the log file using the below configurations in the server where we are running the compaction.

This is one time activity and we need to use the command below to log the information/errors into the log file.

➔ Create the below config file the same place where we have oak-run-*.jar file

➔ Name it as “logback-compaction.xml” and add the below in the lines in xml file.

<appender name="STDERR" class="ch.qos.logback.core.ConsoleAppender">
    <target>System.err</target>
    <encoder>
      <pattern>%d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n</pattern>
    </encoder>
</appender>
<logger name="org.apache.jackrabbit.oak.plugins.segment.Compactor" level="INFO"/>
<root level="warn">
    <appender-ref ref="STDERR" />
</root>
</configuration>

➔ Now run the below command to log the error/info into the log file.

nohup java -Dtar.memoryMapped=true -Dupdate.limit=5000000 -Dcompaction-progress-log=1500000 -Dcompress-interval=10000000 -Doffline-compaction=true -Dlogback.configurationFile=logback-compaction.xml -Xmx10g -jar <oak jar file path> compact <aem-installation-path> > tarcompaction.log 2>&1

➔ Now should be able to see the file name “tarcompaction.log” the place where you created the above xml config.

TAR OFFLINE COMPACTION PREREQUISTICS:

· How to find correct version of oak-run jar?

Go to felix console (/system/console/bundles) –> and search for oak, take version across each oak bundle.

Find Check Points:

----------------------

· java -jar /folderPath/aem/oak-run.jar checkpoints F:/AEM/AEM-4502/crx-quickstart/repository/segmentstore

Run below command on linux/unix machine as windows does not support -Dtar option

· java -Dtar.memoryMapped=true -Xmx8g -jar /folderPath/aem/oak-run-jar/oak-run-1.2.7.jar checkpoints /folderPath/aem/crx-quickstart/repository/segmentstore

Remove CheckPoints:

--------------------------

· java -Dtar.memoryMapped=true -Xmx4g -jar /folderPath/aem/oak-run-jar/oak-run-1.2.7.jar checkpoints /folderPath/aem/crx-quickstart/repository/segmentstore rm-unreferenced

Finally Run Compact:

-------------------------

· java -Dtar.memoryMapped=true -Xmx8g -jar /folderPath/aem/oak-run-jar/oak-run-1.2.7.jar compact /folderPath/aem/crx-quickstart/repository/segmentstore >> /folderPath/aem/help/logs/compactLog

Running Script File:

· Go to respective script file in folder path.

· Running below command will execute all the above 3 commands (i.e. finding checkpoints, deleting checkpoints, compacting) are incorporated with in that script file.

Command:

./scriptFileName

ONLINE COMPACTION:

For situations where the AEM cannot be shut down for maintenance, compaction can also be performed while the instance is running. This is called Online Compaction.

You can configure Online Compaction by doing the following:

➔ Go to the folder where AEM is installed, then browse to crx-quickstart\install

➔ Open the org.apache.jackrabbit.oak.plugins.segment.SegmentNodeStoreService.config file.

➔ If that doesn’t exist create one and add the following line to the configuration file

repository.home=${repository.home}/segmentstore

tarmk.size=256

pauseCompaction=false

➔ Restart AEM

To verify the configuration has taken place, check here.

➔ Go to the JMX console by pointing your browser to

http://localhost:4502/system/console/jmx

➔ Search for CompactionStrategy and click the MBean that shows up in the search.

➔ Next, verify that the value for PausedCompaction is set to false. This confirms that online compaction is set to run.

➔ Next, verify if Online Compaction is running properly. You can do this by first going to the Operations Dashboard and checking what is the time interval configured for the Daily Maintenance Window. By default, it is scheduled to run between 2 and 5 AM.

➔ Now, inspect the error.log file for events logged during the time of the daily maintenance window to see if online compaction ran correctly.

➔ Example Log:

[TarMK compaction thread [/author/crx-quickstart/repository/segmentstore], active since Thu Mar 19 02:00:10 EDT 2015, previous max duration 1369831ms] org.apache.jackrabbit.oak.plugins.segment.file.FileStore TarMK compaction started

19.03.2015 02:00:30.441 *INFO* [pool-9-thread-2] com.adobe.granite.taskmanagement.impl.jcr.TaskArchiveService archiving tasks at: 'Thu Mar 19 02:00:30 EDT 2015'

19.03.2015 02:01:01.699 *INFO* [TarMK compaction thread [/author/crx-quickstart/repository/segmentstore], active since Thu Mar 19 02:00:10 EDT 2015, previous max duration 1369831ms] org.apache.jackrabbit.oak.plugins.segment.file.FileStore Estimated compaction in 51.47 s, gain is 69% (1018859520/3343598080) or (1.0 GB/3.3 GB), so running compaction

➔ Log to make sure online compaction is completed.

How to automate the offline compaction process.

Offline Tar Compaction is still the Adobe recommended way of compacting Oak.

Below is the script which automates entire process.

For above process download a version of Oak Run that matches your repository version.

Steps to follow:

Shutdown AEM

Find Old Checkpoints

Remove Unreferenced Checkpoints

Compact Oak (using compact keyword in command).

Restart AEM

SCRIPT:

#!/bin/bash

todayDate="$(date +'%d-%m-%Y')"

logfile="compact-$ todayDate.log"

installfolder="/data/aem"

aemfolder="$installfolder/crx-quickstart"

oakrun="$installfolder/help/oak-run-1.0.18.jar"

## Shutdown AEM

printf "Shutting down AEM.\n"

$aemfolder/bin/stop

todayDate ="$(date)"

echo "AEM Shutdown at: $ todayDate " >> $installfolder/help/logs/$logfile

## Find old checkpoints

printf "Finding old checkpoints.\n"

java -Dtar.memoryMapped=true -Xms8g -Xmx8g -jar $oakrun checkpoints $aemfolder/repository/segmentstore >> $installfolder/help/logs/$logfile

## Delete unreferenced checkpoints

printf "Deleting unreferenced checkpoints.\n"

java -Dtar.memoryMapped=true -Xms8g -Xmx8g -jar $oakrun checkpoints $aemfolder/repository/segmentstore rm-unreferenced >> $installfolder/help/logs/$logfile

## Run compaction

printf "Running compaction. This may take a while.\n"

java -Dtar.memoryMapped=true -Xms8g -Xmx8g -jar $oakrun compact $aemfolder/repository/segmentstore >> $installfolder/help/logs/$logfile

## Report Completed

printf "Compaction complete. Please check the log at:\n"

printf "$installfolder/help/logs/$logfile\n"

## Start AEM back up

todayDate ="$(date)"

printf "Starting up AEM.\n"

$aemfolder/bin/start

echo "AEM Startup at: $ todayDate " >> $installfolder/help/logs/$logfile

http://experiencemanaged.com/posts/scripting-aem-oak-compaction.html

https://jackrabbit.apache.org/oak/docs/command_line.html