Jupyter/MPI - Revision history

Admin: /* Resources */

2017-11-28T13:12:48Z

Resources

← Older revision		Revision as of 13:12, 28 November 2017
Line 263:		Line 263:
	Notebook to illustrate ipython usage of pxlocal: https://nbviewer.jupyter.org/gist/minrk/4470122		Notebook to illustrate ipython usage of pxlocal: https://nbviewer.jupyter.org/gist/minrk/4470122

			https://github.com/ipython/ipython/pull/8214


	[[Category:Jupyter]]		[[Category:Jupyter]]
	[[Category:Python]]		[[Category:Python]]

2017-11-28T12:58:09Z

Problems sharing a variable using px magic

← Older revision		Revision as of 12:58, 28 November 2017
Line 256:		Line 256:
	* Cell magic: https://ipython.readthedocs.io/en/stable/interactive/magics.html#cell-magics		* Cell magic: https://ipython.readthedocs.io/en/stable/interactive/magics.html#cell-magics
	* All magic: https://ipyparallel.readthedocs.io/en/latest/magics.html		* All magic: https://ipyparallel.readthedocs.io/en/latest/magics.html



			=Resources=

			Notebook to illustrate ipython usage of pxlocal: https://nbviewer.jupyter.org/gist/minrk/4470122

2017-11-28T05:29:06Z

Start cluster with MPI (success)

← Older revision		Revision as of 05:29, 28 November 2017
Line 185:		Line 185:
	===Start cluster with MPI (success)===		===Start cluster with MPI (success)===

	The ~~thing~~ runs when I change to:		The cluster runs when I change to:

	<pre>		<pre>
Line 191:		Line 191:
	</pre>		</pre>

	~~but in the notebook,~~ when I ~~try and create a~~ cluster using		and when I connect to the cluster using:

	<pre>		<pre>
			from ipyparallel import Client, error
	cluster = Client(profile="mpi")		cluster = Client(profile="mpi")
	~~</pre>~~

	~~I get an error message:~~

	~~<pre>~~

	</pre>		</pre>

2017-11-28T05:27:37Z

Start cluster with MPI (success)

← Older revision		Revision as of 05:27, 28 November 2017
Line 188:		Line 188:

	<pre>		<pre>
	ipcluster start -n 2 --engines=MPIEngineSetLauncher		$ ipcluster start --n=2 --engines=MPIEngineSetLauncher --profile=mpi
	</pre>		</pre>

2017-11-28T05:26:01Z

Start cluster with MPI (failures)

← Older revision		Revision as of 05:26, 28 November 2017
Line 84:		Line 84:

	<pre>		<pre>
	vim IPYTHONDIR/profile_mpi/ipcluster_config.py		vim $IPYTHONDIR/profile_mpi/ipcluster_config.py
	</pre>		</pre>

2017-11-28T05:23:04Z

Start cluster with MPI

← Older revision		Revision as of 05:23, 28 November 2017
Line 46:		Line 46:
	</pre>		</pre>

	===Start cluster with MPI===		===Start cluster with MPI (failures)===

	If you do pass an <code>--engines</code> flag, though, it could be problematic:		If you do pass an <code>--engines</code> flag, though, it could be problematic:
Line 183:		Line 183:
	</pre>		</pre>

	~~So the~~ thing runs when I change to:		===Start cluster with MPI (success)===

			The thing runs when I change to:

	<pre>		<pre>
			ipcluster start -n 2 --engines=MPIEngineSetLauncher
	</pre>		</pre>

Line 197:		Line 200:

	<pre>		<pre>
	~~OSError: Connection file '~/.ipython/profile_mpi/security/ipcontroller-client.json' not found.~~
	~~You have attempted to connect to an IPython Cluster but no Controller could be found.~~
	~~Please double-check your configuration and ensure that a cluster is running.~~
	</pre>		</pre>

2017-11-27T16:25:27Z

Start cluster with MPI

@@ Line 181: / Line 181: @@
 -11-27 08:22:00.027 [IPClusterStart] Process '/usr/local/opt/python3/bin/python3.6' stopped: {'exit_code': 1, 'pid': 54085}
 -11-27 08:22:02.910 [IPClusterStart] Removing pid file: /Users/charles/.ipython/profile_default/pid/ipcluster.pid
 </pre>

2017-11-27T16:22:52Z

Start cluster with MPI

@@ Line 107: / Line 107: @@
 This is still not working... more info: https://stackoverflow.com/questions/33614100/setting-up-a-distributed-ipython-ipyparallel-mpi-cluster#33671604
-Thought I just forgot to run a controller, but this doesn't help fix anything:
+<s>Thought I just forgot to run a controller, but this doesn't help fix anything:
 <pre>
 ipcontroller --ip='*' --profile=mpi
 </pre>

2017-11-27T16:19:19Z

Start cluster with MPI

← Older revision		Revision as of 16:19, 27 November 2017
Line 78:		Line 78:

	<pre>		<pre>
	$ ipython profile create --parallel --profile=mpi		ipython profile create --parallel --profile=mpi
	</pre>		</pre>

Line 96:		Line 96:

	<pre>		<pre>
	$ ipcluster start -n 4 --profile=mpi		ipcluster start -n 4 --profile=mpi
	</pre>		</pre>

Line 103:		Line 103:
	<pre>		<pre>
	ipcluster start --engines=MPI --n=4 --log-to-file=True --log-level DEBUG		ipcluster start --engines=MPI --n=4 --log-to-file=True --log-level DEBUG
			</pre>

			This is still not working... more info: https://stackoverflow.com/questions/33614100/setting-up-a-distributed-ipython-ipyparallel-mpi-cluster#33671604

			Thought I just forgot to run a controller, but this doesn't help fix anything:

			<pre>
			ipcontroller --ip='*' --profile=mpi
	</pre>		</pre>

2017-11-27T16:06:24Z

Start cluster with MPI

← Older revision		Revision as of 16:06, 27 November 2017
Line 97:		Line 97:
	<pre>		<pre>
	$ ipcluster start -n 4 --profile=mpi		$ ipcluster start -n 4 --profile=mpi
			</pre>

			If it is still giving you trouble, try dumping debug info:

			<pre>
			ipcluster start --engines=MPI --n=4 --log-to-file=True --log-level DEBUG
	</pre>		</pre>