-
Notifications
You must be signed in to change notification settings - Fork 6
Server installation guide
- Oracle JDK >= 1.5
- Ice 3.3 or 3.4 (version 3.4 is highly recommended)
- make sure you install both Java and Python libraries when installing Ice
- Numpy
- PyTables to enable OMERO.tables
git clone --recursive git://github.com/openmicroscopy/openmicroscopy.git
switch to the master branch:
cd ${OMERO_HOME}
git checkout master
The master branch will usually correspond to the latest stable OMERO release. If you need a specific release, you can switch to it by using the following command:
git reset --hard ${VERSION}
To check all available versions:
git tag -l
Finally, update the submodules:
git submodule update
Refer to the official documentation
Check out the OMERO.biobank source code:
git clone https://github.com/crs4/omero.biobank.git
and copy the custom model files to OMERO's model repository
cp ${BIOBANK_REPO}/models/*.ome.xml ${OMERO_HOME}/components/model/resources/mappings/
Edit the ${OMERO_HOME}/etc/omero.properties file:
omero.db.host=${POSTGRESQL_HOST}
omero.db.name=${OMERO_DB_NAME}
omero.db.user=${OMERO_DB_USER}
omero.db.pass=${OMERO_DB_PASSWD}
You can set the OMERO data directory in ${OMERO_HOME}/etc/omero.properties, but this works only as a reminder. We will soon show how to configure it through the admin shell.
Since OMERO.biobank runs on a modified version of OMERO that uses custom models, a specific tag must be added to the omero.version property. This allows to perform a version check to ensure that the OMERO server supports the specific set of models used by the client. To apply the patch, run:
./utils/patch_ome_config --ome-home ${OMERO_HOME} --models-repository ${MODELS_REPOSITORY}
where ${MODELS_REPOSITORY} can be the models directory of the repository.
In order to handle large datasets, you may need to increase the memory assigned to the Blitz process. Open ${OMERO_HOME}/etc/grid/templates.xml with a text editor, look for the <server-template id="BlitzTemplate"> tag and edit the following lines:
<server-template id="BlitzTemplate">
...
<target name="Blitz-hprof">
...
</target>
<option>-Xmx${MAX_BLITZ_MEM_IN_MB}M</option>
<option>-XX:MaxPermSize=${MAX_PERM_SIZE_IN_MB}m</option>
...
The MaxPermSize parameter is required by Hibernate and represents an extra amount of memory that will be assigned to the Blitz server.
To handle large results from a single query, you need to increase Ice's maximum message size.
In ${OMERO_HOME}/components/blitz/resources/omero/Constants.ice, set:
const int MESSAGESIZEMAX = ${MAX_MESSAGE_SIZE_IN_MB};
In ${OMERO_HOME}/etc/ice.config:
Ice.MessageSizeMax=${MAX_MESSAGE_SIZE_IN_MB}
In ${OMERO_HOME}/etc/grid/templates.xml:
<properties id="Basics">
<property name="Ice.MessageSizeMax" value="${MAX_MESSAGE_SIZE_IN_MB}"/>
...
</properties>
OMERO.biobank can use a graph DB to take advantage of a more reliable and faster lookup engine for the retrieval of the "chain of custody" of the various objects stored in the system. The libraries also need a message engine to synchronize data between the OMERO database and the graph DB.
We use Neo4j as graph DB and RabbitMQ as message engine. To install these tools, refer to the official documentation:
Neo4j and RabbitMQ configuration parameters must be set at the end of $OMERO_HOME/etc/omero.properties (these values will be retrieved when installing the OMERO.biobank API):
...
omero.biobank.graph.engine=neo4j
omero.biobank.graph.uri=${NEO4J_SERVER_URI}
omero.biobank.graph.user=${NEO4J_USER} #optional
omero.biobank.graph.password=${NEO4J_PASSWORD} #optional
omero.biobank.messages_queue.enabled=True
omero.biobank.messages_queue.host=${RABBITMQ_SERVER_HOST}
omero.biobank.messages_queue.port=${RABBITMQ_SERVER_PORT} #optional
omero.biobank.messages_queue.queue=${RABBITMQ_QUEUE_NAME}
omero.biobank.messages_queue.user=${RABBITMQ_USER} #optional
omero.biobank.messages_queue.password=${RABBITMQ_PASSWORD} #optional
NOTE 1: all variables listed as optional must be set only if you change the default Neo4j/RabbitMQ configuration.
NOTE 2: to set the above parameters when the server is already installed, use the omero config command (see below)
If you don't want to enable Neo4j and RabbitMQ (e.g., on a test server), append the following settings to ${OMERO_HOME}/etc/omero.properties:
...
omero.biobank.graph.engine=pygraph
omero.biobank.messages_queue.enabled=False
Build the server:
${OMERO_HOME}/build.py
The build process may fail with the following error:
The system is out of resources.
Consult the following stack trace for details.
java.lang.OutOfMemoryError: GC overhead limit exceeded
[...]
In this case, open ${OME_HOME}/build.py with a text editor and increase the the value for -Djavac.maxmem.default in calculate_memory_args. For instance:
def calculate_memory_args():
return "-Xmx600M -Djavac.maxmem.default=1024M -Djavadoc.maxmem.default=750M -XX:MaxPermSize=256m".split(" ")Build the DB schema from hibernate classes:
${OMERO_HOME}/build.py build-schema -Domero.db.dialect=org.hibernate.dialect.PostgreSQLDialect
Generate the SQL code to initialize the DB:
${OMERO_HOME}/dist/bin/omero db script
The above command will prompt for the omero DB version and patch (press enter to use the suggested default values) and the root password.
Deploy the SQL file to the PostgreSQL server:
psql -h ${POSTGRESQL_HOST} -U ${DB_USER} ${DB_NAME} < ${OMERO_SQL_FILE}
configure the data directory
${OMERO_HOME}/dist/bin/omero config set omero.data.dir ${OMERO_DATA_DIR}
Start the OMERO server with the following command:
${OMERO_HOME}/dist/bin/omero admin start
If you enabled Neo4j, you will also have to start the graph manager, a daemon that updates the Neo4j database according to the RabbitMQ queue status:
${BIOBANK_HOME}/daemons/graph_manager.py
NOTE: to run the daemon, you have to install OMERO.biobank first.