Spark Management Consolidated
A small module that will load as a singleton class object to manage Spark related things.
Installation
Directly via pip on the command line, in a virtualenv:
pip install https://github.com/matz-e/sparkmanager/tarball/masteror for the current user:
pip install --user https://github.com/matz-e/sparkmanager/tarball/masterUsage
The module itself acts as a mediator to Spark:
import sparkmanager as sm
# Create a new application
sm.create("My fancy name",
[("spark.executor.cores", 4), ("spark.executor.memory", "8g")])
data = sm.spark.range(5)
# Will show up in the UI with the name "broadcasting some data"
with sm.jobgroup("broadcasting some data"):
data = sm.broadcast(data.collect())The Spark session can be accessed via sm.spark, the Spark context via sm.sc. Both attributes are instantiated once the create method is called, with the option to call unambiguous methods from both directly via the SparkManager object:
# The following two calls are equivalent
c = sm.parallelize(range(5))
d = sm.sc.parallelize(range(5))
assert c.collect() == d.collect()Cluster support scripts
Environment setup
To create a self-contained Spark environment, the script provided in examples/env.sh can be used. It is currently tuned to the requirements of the bbpviz cluster. A usage example:
SPARK_ROOT=/path/to/my/spark/installation SM_WORKDIR=/path/to/a/work/directory examples/env.shThe working directory will contain:
A Python virtual environment
A basic Spark configuration pointing to directories within the working directory
An environment script to establish the setup
To use the resulting working environment:
. /path/to/a/work/directory/env.shSpark deployment on allocations
Within a cluster allocation, the script sm_cluster can be used to start a Spark cluster. The script will be automatically installed by pip. To use it, pass either a working directory containing an environment or specify them separately:
sm_cluster startup $WORKDIR
sm_cluster startup $WORKDIR /path/to/some/env.shSimilar, to stop a cluster (not necessary with slurm):
sm_cluster shutdown $WORKDIR
sm_cluster shutdown $WORKDIR /path/to/some/env.shSpark applications then can connect to a master found via:
cat $WORKDIR/spark_masterTL;DR on BlueBrain 5
Setup a Spark environment in your current shell, and point WORKDIR to a shared directory. SPARK_HOME needs to be in your environment and point to your Spark installation. By default, only a file with the Spark master and the cluster launch script will be copied to WORKDIR. Then submit a cluster with:
sbatch -A proj16 -t 24:00:00 -N4 --exclusive -C nvme $(which sm_cluster) startup $WORKDIR1 maintainer
Metadata
1 maintainer
Release files for sparkmanager 0.7.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| sparkmanager-0.7.3.tar.gz | 15.5 kB | Without provenance | Details for sparkmanager-0.7.3.tar.gz |
Release files / sparkmanager-0.7.3.tar.gz
| Download URL | sparkmanager-0.7.3.tar.gz |
|---|---|
| Size | 15.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
bf952e31d9fc6c4945613caae64558625b792596985920bcd4c5fa8b73a97a78
|
|
BLAKE2b-256 checksum How to use checksums |
2f638a2a4e0230d133aa5e7d78a88b715c0231cd5784aaf82f67a833cfe6c4f4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/3.1.1 pkginfo/1.5.0.1 requests/2.23.0 setuptools/41.4.0 requests-toolbelt/0.9.1 tqdm/4.43.0 CPython/3.7.4
|
Security for sparkmanager 0.7.3
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
