Table Of Contents
Previous topic
Next topic
Get Seal
Contributors
Seal is developed by:
and GitHub
Seal is developed by:
All Seal programs generally have the following invocation pattern:
seal <program> [options] <input> <output>
Most programs can accept more than one input path (Seqal being the only exception). Also, while some Seal tools have their own specific options, there are a number of Seal options that all programs accept.
Careful how you specify your input and output paths. If you don’t specify a full URI, your path will be interpreted using the Hadoop cluster’s default file system. See the section below on specifying paths to see some examples.
| Meaning | Short Opt | Long Opt |
|---|---|---|
| Number of reduce tasks to use. | -r <INT> | –num-reducers <INT> |
| Override default Seal config file ($HOME/.sealrc) | -sc <FILE> | –seal-config <FILE> |
| Set the value of a property | -D <prop=value> |
In addition, you can use most Hadoop options directly with the Seal tools. This is handy to override the cluster’s default settings.
seal prq -D seal.prq.min-bases-per-read=54 input output
seal seqal -D seal.seqal.remove_unmapped=true --trimq 15 input_1 input_2 output
seal read_sort --num-reducers 96 -D mapred.compress.map.output=true -ann ref.ann input output
If you find you’re setting the same property all the time, you should consider creating a Seal configuration file and saving your settings in it. The Seal tools will automatically load the file on start up. See the section Seal Config File for details.
Ensure the Hadoop environment variables are set (see Installation - Deploying) so that the Seal tools can find the Hadoop cluster configuration and executables.
Here are some examples showing how to specify input and output paths.
| Default file system: | hdfs://hadoop.mycluster.int:9000/ |
| Specified input path: | /user/me/input_data |
| Fully resolved path: | hdfs://hadoop.mycluster.int:9000/user/me/input_data |
The example above specified an absolute path but doesn’t specify a file system. Therefore Seal resolves the path as being on the default Hadoop volume, hdfs://hadoop.mycluster.int:9000/.
| Default file system: | hdfs://hadoop.mycluster.int:9000/ |
| Specified input path: | input_data |
| Fully resolved path: | hdfs://hadoop.mycluster.int:9000/user/me/input_data |
This example uses a relative path, input_data. It is interpreted as being on the default file system (hdfs://hadoop.mycluster.int:9000/) and in the user’s home directory for that file system. In HDFS, the user’s home directory is always /user/${USER} (the ${USER} variable expands to the current user’s Unix username).
| Default file system: | hdfs://hadoop.mycluster.int:9000/ |
| Specified input path: | file:///home/me/input_data |
| Fully resolved path: | /home/me/input_data |
In this case we see an input path specified as a full URI; in this way we can access data that is not on Hadoop cluster’s default file system. In fact, noticed how the default file system is hdfs://hadoop.mycluster.int:9000/, but we’re accessing data that resides on a standard file system path. Note that for this to work the specified path must exist on all the Hadoop cluster’s nodes.
| Default file system: | file:// |
| Specified input path: | /home/me/input_data |
| Fully resolved path: | /home/me/input_data |
In this example Hadoop has been configured with the default file system being the standard local file system. The input path specified, which is not a full URI, is therefore interpreted as being on the local file system.