[jira] [Commented] (SPARK-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15188589#comment-15188589
]
Eran Withana commented on SPARK-12345:
--
is the resolution to this issue available in Spark 1.6.0 release?
I just used Spark 1.6.0 and got the following error in mesos logs, when it
tried to run the task
```
I0310 03:13:11.417009 131594 exec.cpp:132] Version: 0.23.1
I0310 03:13:11.419452 131601 exec.cpp:206] Executor registered on slave
20160223-000314-3439362570-5050-631-S0
sh: 1: /usr/spark-1.6.0-bin-hadoop2.6/bin/spark-class: not found
```
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Timothy Chen
>Priority: Critical
> Fix For: 1.6.0
>
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15065047#comment-15065047
]
Andrew Or commented on SPARK-12345:
---
For those who are following:
There are 4 patches related to this issue that are merged in this order:
(1) https://github.com/apache/spark/pull/10332 - doesn't actually work
(2) https://github.com/apache/spark/pull/10359 - fixes #10332 to make it
actually work
(3) https://github.com/apache/spark/pull/10366 - fixes #10359, which broke HA
(SPARK-12413)
(4) https://github.com/apache/spark/pull/10329 - an alternative, more correct
fix
Patches (1), (2), and (3) are merged ONLY into branch-1.6.
Patch (4) is merged ONLY in master.
We have a different fix for branch-1.6 because it was an RC blocker and we
wanted to minimize the scope of the changes there. However, patch (4) is a
better fix, and so it exists in master for the longer term.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Timothy Chen
>Priority: Critical
> Fix For: 1.6.0
>
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15060475#comment-15060475
]
Apache Spark commented on SPARK-12345:
--
User 'tnachen' has created a pull request for this issue:
https://github.com/apache/spark/pull/10332
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059937#comment-15059937
]
Saisai Shao commented on SPARK-12345:
-
I think by default Spark Mesos implementation will ship all the environment
variables to the remote nodes, which includes {{SPARK_HOME}}, and Mesos itself
will invoke the spark application through scripts, and inside the scripts we
honor if {{SPARK_HOME}} is already set, so that's the problem.
Basically, I think there're two sides we could fix:
1. We should not expose {{SPARK_HOME}} to the environment if it is not set
specifically. Otherwise cases like here will potentially have problem.
2. Spark on Mesos should not blindly ship all the environment variables to the
remote side. The best way for Spark on Mesos is to invoke the Java program like
what YARN did currently, not rely on scripts.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059954#comment-15059954
]
Saisai Shao commented on SPARK-12345:
-
Having a quick test by not exporting {{SPARK_HOME}}, the application is failed
to start, code in {{SparkLaucher}} needs {{SPARK_HOME}}. So solution 2 is the
only choice to filter out {{SPARK_HOME}} if necessary.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15060035#comment-15060035
]
Saisai Shao commented on SPARK-12345:
-
Here is the one solution
(https://github.com/apache/spark/compare/master...jerryshao:SPARK-12345), mind
taking a trying in your cluster? Thanks a lot.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15060153#comment-15060153
]
Luc Bourlier commented on SPARK-12345:
--
I have almost the same fix, which is the same logic: do not carry `SPARK_HOME`
information across systems. But I changed it in SparkSubmit side:
https://github.com/skyluc/spark/commit/5b6eaa5bf936ef42d46b53564816d62b2aa44e86
I'm running tests to check that Mesos is working fine with those changes.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15060239#comment-15060239
]
Luc Bourlier commented on SPARK-12345:
--
I tested our usual test cases with my change, and it is working well.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15060236#comment-15060236
]
Iulian Dragos commented on SPARK-12345:
---
I'd prefer filtering it at at the submit side, if everything else works. Doing
it in the scheduler will be confusing for users. The Driver environment would
still show SPARK_HOME (in the dispatcher UI), but in fact it would be filtered
out in practice.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059769#comment-15059769
]
Stavros Kontopoulos commented on SPARK-12345:
-
Adding to [~dragos].(MesosClusterScheduler.scala) ...
Mesos Dispatcher passed that env variable to the driver at line:
builder.setEnvironment(envBuilder.build())
also having set executor uri it picks this path:
else if (executorUri.isDefined) {
In 1.6 the spark-submit script has changed...
https://github.com/apache/ spark/blob/master/bin/spark-submit
This line make SPARK_HOME not to change to the local path on mesos slave (from
which the spark-submit was called from) because it has already been set.
if [ -z "${SPARK_HOME}" ]; then
export SPARK_HOME="$(cd "dirname "$0""/..; pwd)"
fi
And the driver is started with spark-submit command anyway...
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059717#comment-15059717
]
Iulian Dragos commented on SPARK-12345:
---
There's isn't any {{SPARK_HOME}} set on any of the Mesos slaves.
Here's what I think happens: the {{SPARK_HOME}} variable is exported by
{{spark-submit}}, and copied
[here|https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/scheduler/cluster/mesos/MesosClusterScheduler.scala#L369-L372]
to the driver environment.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059778#comment-15059778
]
Saisai Shao commented on SPARK-12345:
-
A simple solution is to change the scripts to not expose `SPARK_HOME`. From my
understanding a good solution for Mesos is to do the same way as YARN to
directly invoke Java program, not relying on script to start program.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059812#comment-15059812
]
Sean Owen commented on SPARK-12345:
---
Yeah, but why is {{SPARK_HOME}} copied across machines to begin with? that
seems like the more fundamental issue.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15058873#comment-15058873
]
Iulian Dragos commented on SPARK-12345:
---
[~skonto] pointed out this commit:
https://github.com/apache/spark/commit/8aff36e91de0fee2f3f56c6d240bb203b5bb48ba
it could be part of the problem.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15058887#comment-15058887
]
Apache Spark commented on SPARK-12345:
--
User 'dragos' has created a pull request for this issue:
https://github.com/apache/spark/pull/10315
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059569#comment-15059569
]
Saisai Shao commented on SPARK-12345:
-
It is OK in my local test when I followed the step one by one in doc to run
Spark on Mesos example, though my environment may not be the same as real
cluster.
So I'm not sure how to reproduce this issue, you mean it is introduced by
SPARK-2960, but the behavior of SPARK-2960 is no difference to the previous
version if SPARK_HOME is not manually specified.
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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-12345) Mesos cluster mode is broken
[
https://issues.apache.org/jira/browse/SPARK-12345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15059648#comment-15059648
]
Sean Owen commented on SPARK-12345:
---
Agree, there _shouldn't_ be any change in behavior if {{SPARK_HOME}} is not
set; the change was to use this value if set to find Spark files and scripts.
If unset, it also sets {{SPARK_HOME}} to the location it had inferred before.
However it also seems like there could well be a connection here.
[~dragos] is it possible the cluster has set {{SPARK_HOME}} "incorrectly" or
otherwise copies the value that the driver uses somehow?
> Mesos cluster mode is broken
>
>
> Key: SPARK-12345
> URL: https://issues.apache.org/jira/browse/SPARK-12345
> Project: Spark
> Issue Type: Bug
> Components: Mesos
>Affects Versions: 1.6.0
>Reporter: Andrew Or
>Assignee: Apache Spark
>Priority: Critical
>
> The same setup worked in 1.5.2 but is now failing for 1.6.0-RC2.
> The driver is confused about where SPARK_HOME is. It resolves
> `mesos.executor.uri` or `spark.mesos.executor.home` relative to the
> filesystem where the driver runs, which is wrong.
> {code}
> I1215 15:00:39.411212 28032 exec.cpp:134] Version: 0.25.0
> I1215 15:00:39.413512 28037 exec.cpp:208] Executor registered on slave
> 130bdc39-44e7-4256-8c22-602040d337f1-S1
> bin/spark-submit: line 27:
> /Users/dragos/workspace/Spark/dev/rc-tests/spark-1.6.0-bin-hadoop2.6/bin/spark-class:
> No such file or directory
> {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
