[jira] [Commented] (SPARK-10635) pyspark - running on a different host
[
https://issues.apache.org/jira/browse/SPARK-10635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15579914#comment-15579914
]
holdenk commented on SPARK-10635:
-
it would be a bit difficult, although as Py4J is speeding up the bridge it
might not be impossible. I think realistically though this is probably a won't
fix unless since there doesn't seem to be a lot of interest in this feature.
> pyspark - running on a different host
> -
>
> Key: SPARK-10635
> URL: https://issues.apache.org/jira/browse/SPARK-10635
> Project: Spark
> Issue Type: Improvement
> Components: PySpark
>Reporter: Ben Duffield
>
> At various points we assume we only ever talk to a driver on the same host.
> e.g.
> https://github.com/apache/spark/blob/v1.4.1/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala#L615
> We use pyspark to connect to an existing driver (i.e. do not let pyspark
> launch the driver itself, but instead construct the SparkContext with the
> gateway and jsc arguments.
> There are a few reasons for this, but essentially it's to allow more
> flexibility when running in AWS.
> Before 1.3.1 we were able to monkeypatch around this:
> {code}
> def _load_from_socket(port, serializer):
> sock = socket.socket()
> sock.settimeout(3)
> try:
> sock.connect((host, port))
> rf = sock.makefile("rb", 65536)
> for item in serializer.load_stream(rf):
> yield item
> finally:
> sock.close()
> pyspark.rdd._load_from_socket = _load_from_socket
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-10635) pyspark - running on a different host
[
https://issues.apache.org/jira/browse/SPARK-10635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14908738#comment-14908738
]
Ben Duffield commented on SPARK-10635:
--
Ok good flag that there are other places this'd need to be considered.
How open would you be to a PR which addresses this? I.e. sure it's an
assumption now - could we move away from that?
> pyspark - running on a different host
> -
>
> Key: SPARK-10635
> URL: https://issues.apache.org/jira/browse/SPARK-10635
> Project: Spark
> Issue Type: Improvement
> Components: PySpark
>Reporter: Ben Duffield
>
> At various points we assume we only ever talk to a driver on the same host.
> e.g.
> https://github.com/apache/spark/blob/v1.4.1/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala#L615
> We use pyspark to connect to an existing driver (i.e. do not let pyspark
> launch the driver itself, but instead construct the SparkContext with the
> gateway and jsc arguments.
> There are a few reasons for this, but essentially it's to allow more
> flexibility when running in AWS.
> Before 1.3.1 we were able to monkeypatch around this:
> {code}
> def _load_from_socket(port, serializer):
> sock = socket.socket()
> sock.settimeout(3)
> try:
> sock.connect((host, port))
> rf = sock.makefile("rb", 65536)
> for item in serializer.load_stream(rf):
> yield item
> finally:
> sock.close()
> pyspark.rdd._load_from_socket = _load_from_socket
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-10635) pyspark - running on a different host
[
https://issues.apache.org/jira/browse/SPARK-10635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14802950#comment-14802950
]
Ben Duffield commented on SPARK-10635:
--
Curious as to why you believe this to be hard to support?
We've been using this at many places for quite a long time without issue prior
to 1.4.
I guess there's the question of how to plumb the correct host to
_load_from_socket. I'm also not aware of the reasons of changing the
ServerSocket in python rdd serveIterator to listen explicitly on localhost.
These are the only two places though I believe need to change.
The alternative is for us to put proxying into the application itself (the
application acting as driver) and then monkeypatching pyspark as before, but
this isn't ideal.
> pyspark - running on a different host
> -
>
> Key: SPARK-10635
> URL: https://issues.apache.org/jira/browse/SPARK-10635
> Project: Spark
> Issue Type: Improvement
> Components: PySpark
>Reporter: Ben Duffield
>
> At various points we assume we only ever talk to a driver on the same host.
> e.g.
> https://github.com/apache/spark/blob/v1.4.1/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala#L615
> We use pyspark to connect to an existing driver (i.e. do not let pyspark
> launch the driver itself, but instead construct the SparkContext with the
> gateway and jsc arguments.
> There are a few reasons for this, but essentially it's to allow more
> flexibility when running in AWS.
> Before 1.3.1 we were able to monkeypatch around this:
> {code}
> def _load_from_socket(port, serializer):
> sock = socket.socket()
> sock.settimeout(3)
> try:
> sock.connect((host, port))
> rf = sock.makefile("rb", 65536)
> for item in serializer.load_stream(rf):
> yield item
> finally:
> sock.close()
> pyspark.rdd._load_from_socket = _load_from_socket
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-10635) pyspark - running on a different host
[
https://issues.apache.org/jira/browse/SPARK-10635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14802958#comment-14802958
]
Patrick Woody commented on SPARK-10635:
---
For a bit of motivation - we have a long running SparkContext that essentially
acts as a query server with many clients in iPython Notebook.
We want to keep the driver on a different box from the python kernels to
protect it from potentially resource-heavy python processes (we've had OOM
killer issues in the past). It seems reasonable via py4j, but we are running
into the above issues post-1.4.
> pyspark - running on a different host
> -
>
> Key: SPARK-10635
> URL: https://issues.apache.org/jira/browse/SPARK-10635
> Project: Spark
> Issue Type: Improvement
> Components: PySpark
>Reporter: Ben Duffield
>
> At various points we assume we only ever talk to a driver on the same host.
> e.g.
> https://github.com/apache/spark/blob/v1.4.1/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala#L615
> We use pyspark to connect to an existing driver (i.e. do not let pyspark
> launch the driver itself, but instead construct the SparkContext with the
> gateway and jsc arguments.
> There are a few reasons for this, but essentially it's to allow more
> flexibility when running in AWS.
> Before 1.3.1 we were able to monkeypatch around this:
> {code}
> def _load_from_socket(port, serializer):
> sock = socket.socket()
> sock.settimeout(3)
> try:
> sock.connect((host, port))
> rf = sock.makefile("rb", 65536)
> for item in serializer.load_stream(rf):
> yield item
> finally:
> sock.close()
> pyspark.rdd._load_from_socket = _load_from_socket
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-10635) pyspark - running on a different host
[
https://issues.apache.org/jira/browse/SPARK-10635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14791389#comment-14791389
]
Josh Rosen commented on SPARK-10635:
[~davies], do you think we should support this? This seems like a
hard-to-support feature, so I'm inclined to say that this issue is "Won't Fix"
as currently described.
> pyspark - running on a different host
> -
>
> Key: SPARK-10635
> URL: https://issues.apache.org/jira/browse/SPARK-10635
> Project: Spark
> Issue Type: Improvement
>Reporter: Ben Duffield
>
> At various points we assume we only ever talk to a driver on the same host.
> e.g.
> https://github.com/apache/spark/blob/v1.4.1/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala#L615
> We use pyspark to connect to an existing driver (i.e. do not let pyspark
> launch the driver itself, but instead construct the SparkContext with the
> gateway and jsc arguments.
> There are a few reasons for this, but essentially it's to allow more
> flexibility when running in AWS.
> Before 1.3.1 we were able to monkeypatch around this:
> {code}
> def _load_from_socket(port, serializer):
> sock = socket.socket()
> sock.settimeout(3)
> try:
> sock.connect((host, port))
> rf = sock.makefile("rb", 65536)
> for item in serializer.load_stream(rf):
> yield item
> finally:
> sock.close()
> pyspark.rdd._load_from_socket = _load_from_socket
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
