Hannibal is a tool to help monitor and maintain HBase-Clusters that are configured for manual splitting.

While HBase provides metrics to monitor overall cluster health via JMX or Ganglia, it lacks the ability to monitor single regions in an easy way. This information is essential when your cluster is configured for manual splits, especially when the data growth is not uniform.

This tool tries to fill that gap by answering the following questions:

• How well are regions balanced over the cluster?
• How well are the regions split for each table?
• How do regions evolve over time?

Java 6 JDK is required on the machine where this tool is built.

You will also need a browser with SVG-Support to display Hannibal's graphs.

Hannibal currently supports HBase versions 0.90 to 0.98. The Scala-sources are currently compiled with Apache HBase versions wherever possible, you can try to alter the version in project/Build.scala if you wish to.

1. Download and extract the package according to your HBase version. Grab the Download-URL for the correct HBase-Version from the Release-Page and extract it:

   $ wget <URL-TO-PACKAGE>
   $ tar -xf "hannibal-hbase*.tgz"
   $ cd hannibal
   

2. Copy conf/hbase-site.template.xml to conf/hbase-site.xml and adjust it:

   $ cp conf/hbase-site.template.xml conf/hbase-site.xml
   $ vi conf/hbase-site.xml
   

3. Run the start script inside the root folder of the project:

   $ ./start
   

When the application has started, you can access the web UI at: http://localhost:9000

Please note that history data about regions is only collected while the application is running, it will need to run for some time until the region detail graphs fill up.

For information about the usage, check out the Usage page on our Wiki.

1. Grab the sources from github:

   $ git clone https://github.com/sentric/hannibal.git
   $ cd hannibal
   

2. Set the Environmentvariable HANNIBAL_HBASE_VERSION according to your HBase version. For example for HBase 0.98 do:

   $ export HANNIBAL_HBASE_VERSION=0.98
   

   Other possible values are "0.90", "0.94", "0.96" or "0.98". Be sure to always have this environment-variable set before executing build or start scripts.

3. Copy conf/hbase-site.template.xml to conf/hbase-site.xml and adjust it.

4. Build the project using the build script inside the root folder of the project:

   $ ./build
   

5. Run the start script inside the root folder of the project:

   $ ./start
   

The first time you build, sbt will fetch all dependencies needed to build and run the the application. It will take a few minutes to build.

When the application has started, you can access the web UI at: http://localhost:9000

Please note that history data about regions is only collected while the application is running, it will need to run for some time until the region detail graphs fill up.

For information about the usage, check out the Usage page on our Wiki.

There is also a tutorial video on [YouTube][yt], that shows how to install and use Hannibal for HBase 0.90. [yt]: http://www.youtube.com/watch?v=gu0lGAf7JO8

There is also Hue App to integrate Hannibal into Hue. It is available here.

If you have an Android Device, you might be interested in the Mobile App. The App which can be downloaded from Google Play. Further details can be found in the Wiki.

If you want to make Hannibal reachable from the Internet, it's recommended to put a web server in front of it that takes care of handling authentication. Instructions can be found in the Wiki.

HBase 0.90.x's API doesn't allow you to query information on running compactions directly, so what we do is to parse the RegionServers' log files directly, which are available through the service interface. HBase 0.92 and up allows you to query compactions directly, but we still collect compactions using the logfile-parsing technique, because this way we don't miss any short running compactions. The downside is that this doesn't work out of the box for all HBase clusters because either, the path-pattern or the date-pattern can differ from system to system. Another problem can be, that the compaction-information isn't logged at all in your setup, because your LogLevel is set too high.

If you run into problems with the the compaction-metrics, you should check the following parameters in conf/application.conf.

The default of the logfile.path-pattern is

logfile.path-pattern = "(?i)\"/logs/(.*regionserver.*[.].*)\""

The defaults should work for most setups in distributed mode. For standalone mode you will need change the pattern to

logfile.path-pattern = (?i)\"/logs/(.*master.*[.].*)\"

If you are still unsure about the correct path-pattern, you can get a hint for the correct pattern by looking at your log-listing http://<<some-regionserver>>:60030/logs/.

The default logfile.date-format is

logfile.date-format = "yyyy-MM-dd HH:mm:ss,SSS"

You can figure out the correct date-format by looking inside a logfile within your log-listing at http://<<some-regionserver>>:60030/logs/

Informations about compactions are logged by HBase with INFO-Level, so the log-level for your HBase-Regionservers need to be set at least to INFO.

Hannibal can set the log level to INFO for you, just edit conf/application.conf and set

logfile.set-loglevels-on-startup = true

If this doesn't work for you, you should try to manually change the loglevel on your regionservers.

Hannibal has been already been configured to look for the INFO message containing compaction metrics (with some default regex patterns depending on the version of hbase.) Some times the INFO message logged might not match with the default regex pattern(maybe its different distribution of base), which would result in you not being able to see the compaction metrics.

if this is the case, you can set the regex pattern via this variable example: logfileParser.overrideCompactionRegexPattern="""^(.) INFO (.).CompactionRequest: completed compaction: regionName=(..), storeName=(.), fileCount=(.), fileSize=(.), priority=(.), time=(.); duration=(.*)$"""

Also set the positions of the date/region/duration groups in the above regex via the following parameters:

logfileParser.dateGroupPosition=1 logfileParser.regionGroupPosition=3 logfileParser.durationGroupPosition=9

i.e. date occurs as the first group, region as the third group and duration as the ninth group in the above example regex pattern

Hannibal is not yet ready to be used on large clusters (say about more than 100 machines), however some work has been done (thanks to Alexandre Normand and churrodog) to make it at least possible to run it without crashing on mid-sized clusters. If Hannibal's performance is not sufficient with your HBase setup, it may help to tune the following parameters in conf/application.conf.

This defaults to 300 seconds and it defines how often logfiles are collected from the different region servers. This process is quite heavy and for big clusters you should consider either to increase the interval or to disable it altogether by setting it to 0

metrics.logfile-fetch-interval = 0 # Disables compaction-metrics

This defaults to 60 seconds and you usually want to keep the update-interval high, because this way the UI will present you the most recent values and also will the region-history graph. However if you have many regions and it takes long to update the region metrics you may set this one to a value up to about 1800 (= 30 Minutes).

metrics.regions-fetch-interval = 600 # 10 Minutes

Old metrics are cleaned after one day by default and it makes sense since the region history graphs are presenting data up to one day. However if you have a very large number of regions and your H2 database just explodes you may consider to reduce them.

If you intend to run Hannibal on a different host from where you want to build it, then you can run

./create_package

This script generates a tgz-package inside the target folder, which you can then deploy on your target server. The HBase version can be set with the HANNIBAL_HBASE_VERSION environment variable, as described in the quickstart section.

More information about the tool can be found in the Wiki

Hannibal is released under MIT License, see LICENSE for details.

If you need additional features or help please get in touch with us. Subscribe to the Hannibal mailing list or follow Hannibal App on Twitter.

This tool was developed at Sentric by Nils Kübler and Jiayong Ou. Sentric has been acquired by YMC AG in May 2013.

With help from:

• Jean-Pierre König
• Christian Gügi
• Vadim Kisselmann
• Ben Taylor
• Stephanie Höhn
• Alexandre Normand
• churrodog