Configuring Pulsar¶
If either installation procedure has been followed, your Pulsar directory
should contain two files of interest: app.yml
to configure the Pulsar
application and server.ini
to configure the web server (unless you are
running Pulsar without a web server).
Default values are specified for all configuration options that will work if
Pulsar is running on the same host as Galaxy (e.g. for testing and development).
Otherwise, the host
setting of server.ini
will need to be modified to
listen for external requests.
app.yml
settings can be overridden by
setting environment variables, just as with Galaxy, by prefixing the config
setting name with PULSAR_CONFIG_OVERRIDE_
. For example:
$ export PULSAR_CONFIG_OVERRIDE_PRIVATE_TOKEN=changed
$ pulsar
Defaults can also be set via environment variables by prefixing them with
PULSAR_CONFIG_
. For example, PULSAR_CONFIG_PRIVATE_TOKEN
.
Security¶
Out of the box, Pulsar essentially allows anyone with network access to the Pulsar server to execute arbitrary code and read and write any files the web server can access. Hence, in most settings steps should be taken to secure the Pulsar server.
Private Token¶
If running Pulsar with a web server, you must specify a private token (a
shared secret between Pulsar and the Galaxy server) to prevent unauthorized
access. This is done by simply setting private_token
in app.yml
to some
long random string.
Once a private token is configured, Galaxy job destinations should include a
private_token
parameter to authenticate these jobs.
Pulsar Web Server¶
The default Pulsar web server, Paste can be configured to use SSL and to require the client (i.e. Galaxy) to pass along a private token authorizing use.
Tip
SSL support is built in to uWSGI, an alternate webserver that can be installed (see Installing Pulsar).
pyOpenSSL
is required to configure a Pulsar web server to server content via
HTTPS/SSL. This dependency can be difficult to install and seems to be getting
more difficult. Under Linux you will want to ensure the needed dependencies to
compile pyOpenSSL are available - for instance in a fresh Ubuntu image you
will likely need:
$ sudo apt-get install libffi-dev python3-dev libssl-dev
Then pyOpenSSL can be installed with the following command (be sure to source your virtualenv if setup above):
$ pip install pyOpenSSL
Once installed, you will need to set the option ssl_pem
in server.ini
.
This parameter should reference an OpenSSL certificate file for use by the
Paste server. This parameter can be set to *
to automatically generate such
a certificate. An unsigned certificate for testing purposes can be manually
generated by the following method:
$ openssl genrsa 1024 > host.key
$ chmod 400 host.key
$ openssl req -new -x509 -nodes -sha1 -days 365 \
-key host.key > host.cert
$ cat host.cert host.key > host.pem
$ chmod 400 host.pem
More information can be found in the paste httpserver documentation.
Message Queue¶
If Pulsar is processing requests via a message queue instead of a web server the underlying security
mechanisms of the message queue should be used to secure communication -
deploying Pulsar with SSL and a private_token
described above are not
applicable.
This can be done via two (not mutually exclusive) methods: client SSL certificates, or password authentication. In either case, you should configure your AMQP server with SSL.
If using client certificates, you will likely need to set the appropriate (for
your PKI) combination of amqp_connect_ssl_ca_certs
,
amqp_connect_ssl_keyfile
, amqp_connect_ssl_certfile
, and
amqp_connect_ssl_cert_reqs
, in Pulsar’s app.yml
file. See
app.yml.sample
for more details.
If using password authentication, this information can be set in the
message_queue_url
setting in app.yml
, e.g., with SSL:
message_queue_url: amqps://user:password@mqserver.example.org:5671//
You can consult the Kombu documentation for even more information.
User Authentication/Authorization¶
You can configure Pulsar to authenticate user during request processing and check if this user is allowed to run a job.
Various authentication/authorization plugins can be configured in app.yml to do that and plugin parameters depend on auth type. For example, the following configuration uses oidc plugin for authentication and userlist for authorization:
user_auth:
authentication:
- type: oidc
oidc_jwks_url: https://login.microsoftonline.com/xxx/discovery/v2.0/keys
oidc_provider: azure
oidc_username_in_token: preferred_username
oidc_username_template: *.
authorization:
- type: userlist
userlist_allowed_users:
- xxx
see plugins folder for available plugins and their parameters.
Customizing the Pulsar Environment (*nix only)¶
For many deployments, Pulsar’s environment will need to be tweaked. For
instance to define a DRMAA_LIBRARY_PATH
environment variable for the
drmaa
Python module or to define the location to a find a location of
Galaxy (via GALAXY_HOME
) if certain Galaxy tools require it or if Galaxy
metadata is being set by the Pulsar.
The file local_env.sh
(created automatically by pulsar-config
) will be
source by pulsar
before launching the application and by child process
created by Pulsar that require this configuration.
Job Managers (Queues)¶
By default the Pulsar will maintain its own queue of jobs. While ideal for simple deployments such as those targeting a single Windows instance, if Pulsar is going to be used on more sophisticated clusters, it can be configured to maintain multiple such queues with different properties or to delegate to external job queues (via DRMAA, qsub/qstat CLI commands, or Condor).
For more information on configured external job managers, see Job Managers.
Galaxy Tools¶
Some Galaxy tool wrappers require a copy of the Galaxy codebase itself to run.
Such tools will not run under Windows, but on *nix hosts the Pulsar can be
configured to add the required Galaxy code a jobs PYTHON_PATH
by setting
GALAXY_HOME
environment variable in the Pulsar’s local_env.sh
file
(described above).
Most Galaxy tools require external command-line tools, known as Galaxy Tool
Dependencies, to execute correctly. In Galaxy, these are provided by its
Dependency Resolution system. Pulsar uses this same system, which can be
configured via the dependency_resolution
option in app.yml
. See the
example in app.yml.sample for additional information. In its default
configuration, Pulsar will automatically install Conda but not automatically
install missing tool dependencies. Administrators sending large numbers of tools
to Pulsar most likely want to enable the auto_install
option on the
conda
dependency resolver or the conda_auto_install
global option so
that it is not necessary to manually install dependencies for tools sent to
Pulsar. Both options are documented in the app.yml.sample file.
Message Queue (AMQP)¶
Galaxy and Pulsar can be configured to communicate via a message queue instead of a Pulsar web server. In this mode, Pulsar and Galaxy will send and receive job control and status messages via an external message queue server using the AMQP protocol. This is sometimes referred to as running Pulsar “webless”.
Information on configuring RabbitMQ, one such compatible message queue, can be found in Message Queues with Galaxy and Pulsar.
In addition, when using a message queue, Pulsar will download files from and upload files to Galaxy instead of the inverse. Message queue mode may be very advantageous if Pulsar needs to be deployed behind a firewall or if the Galaxy server is already set up (via proxy web server) for large file transfers.
A template configuration for using Galaxy with a message queue can be created
by pulsar-config
:
$ pulsar-config --mq
You will also need to ensure that the ` kombu
Python dependency is installed
(pip install kombu
). Once this is available, simply set the
message_queue_url
property in app.yml
to the correct URL of your
configured AMQP endpoint.
AMQP does not guarantee message receipt. It is possible to have Pulsar (and
Galaxy) require acknowledgement of receipt and resend messages that have not
been acknowledged, using the amqp_ack*
options documented in
app.yml.sample, but beware that enabling this option can give rise to the
Two Generals Problem, especially when Galaxy or the Pulsar server are down
(and thus not draining the message queue).
In the event that the connection to the AMQP server is lost during message
publish, the Pulsar server can retry the connection, governed by the
amqp_publish*
options documented in app.yml.sample.
Caching (Experimental)¶
Pulsar and its client can be configured to cache job input files. For some
workflows this can result in a significant decrease in data transfer and
greater throughput. On the Pulsar server side - the property
file_cache_dir
in app.yml
must be set. See Galaxy’s job_conf.xml example file for information on configuring the client.
More discussion on this can be found in this galaxy-dev mailing list thread and future plans and progress can be tracked on this Trello card.