spark-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dong Lei <dong...@microsoft.com>
Subject RE: How to support dependency jars and files on HDFS in standalone cluster mode?
Date Thu, 11 Jun 2015 06:40:38 GMT
I think in standalone cluster mode, spark is supposed to do:

1.       Download jars, files to driver

2.       Set the driver’s class path

3.       Driver setup a http file server to distribute these files

4.       Worker download from driver and setup classpath

Right?

But somehow, the first step fails.
Even if I can make the first step works(use option1), it seems that the classpath in driver
is not correctly set.

Thanks
Dong Lei

From: Cheng Lian [mailto:lian.cs.zju@gmail.com]
Sent: Thursday, June 11, 2015 2:32 PM
To: Dong Lei
Cc: Dianfei (Keith) Han; dev@spark.apache.org
Subject: Re: How to support dependency jars and files on HDFS in standalone cluster mode?

Oh sorry, I mistook --jars for --files. Yeah, for jars we need to add them to classpath, which
is different from regular files.

Cheng
On 6/11/15 2:18 PM, Dong Lei wrote:
Thanks Cheng,

If I do not use --jars how can I tell spark to search the jars(and files) on HDFS?

Do you mean the driver will not need to setup a HTTP file server for this scenario and the
worker will fetch the jars and files from HDFS?

Thanks
Dong Lei

From: Cheng Lian [mailto:lian.cs.zju@gmail.com]
Sent: Thursday, June 11, 2015 12:50 PM
To: Dong Lei; dev@spark.apache.org<mailto:dev@spark.apache.org>
Cc: Dianfei (Keith) Han
Subject: Re: How to support dependency jars and files on HDFS in standalone cluster mode?

Since the jars are already on HDFS, you can access them directly in your Spark application
without using --jars

Cheng
On 6/11/15 11:04 AM, Dong Lei wrote:
Hi spark-dev:

I can not use a hdfs location for the “--jars” or “--files” option when doing a spark-submit
in a standalone cluster mode. For example:
                Spark-submit  …   --jars hdfs://ip/1.jar  ….  hdfs://ip/app.jar (standalone
cluster mode)
will not download 1.jar to driver’s http file server(but the app.jar will be downloaded
to the driver’s dir).

I figure out the reason spark not downloading the jars is that when doing sc.addJar to http
file server, the function called is Files.copy which does not support a remote location.
And I think if spark can download the jars and add them to http file server, the classpath
is not correctly set, because the classpath contains remote location.

So I’m trying to make it work and come up with two options, but neither of them seem to
be elegant, and I want to hear your advices:

Option 1:
Modify HTTPFileServer.addFileToDir, let it recognize a “hdfs” prefix.

This is not good because I think it breaks the scope of http file server.

Option 2:
Modify DriverRunner.downloadUserJar, let it download all the “--jars” and “--files”
with the application jar.

This sounds more reasonable that option 1 for downloading files. But this way I need to read
the “spark.jars” and “spark.files” on downloadUserJar or DriverRunnder.start and replace
it with a local path. How can I do that?


Do you have a more elegant solution, or do we have a plan to support it in the furture?

Thanks
Dong Lei


Mime
View raw message