You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@oozie.apache.org by Jay Vyas <ja...@gmail.com> on 2014/02/13 20:47:51 UTC
oozie java action : Why is the nameNode required?
Is a nameNode a required element for plain java or mapreduce oozie actions?
It seems so, but im not sure why:
oozie job -oozie http://localhost:11000/oozie -config job.properties -run
Error: E0701 : E0701: XML schema error, cvc-complex-type.2.4.a: Invalid
content was found starting with element 'configuration'. One of
'{"uri:oozie:workflow:0.1":name-node}' is expected.
It seems like the namenode should be an optional parameter, since according
to https://issues.apache.org/jira/browse/OOZIE-426, OOZIE is written
directly against the file system API, without any coupling to HDFS/.
--
Jay Vyas
http://jayunit100.blogspot.com
Re: Oozie permissions, task controllers, appTokens
Posted by Jay Vyas <ja...@gmail.com>.
Thanks again mona:
1) Not sure what the purpose of the auth.token is - is that required for
running Linux TaskControllers?
2) Is it customary to submit jobs as user "oozie"?
Is there a doc on this somewhere ? Im not an oozie expert : just want my
user "tom" to be able to submit jobs that the LinuxTaskControllers (running
as "tom") can see - im afraid though that if i run "oozie", it might not
work properly with the yarn LXC .
Anyways, let me know if im missing something.
Re: Oozie permissions, task controllers, appTokens
Posted by Mona Chitnis <ch...@yahoo-inc.com>.
The permissions for the .oozie.auth.token are 600. If you are submitting
Oozie jobs as a different user (task controller) than yourself, consider
authorizing as that user first.
On 2/13/14, 7:10 PM, "Jay Vyas" <ja...@gmail.com> wrote:
>What is the permissions model for appTokens and other job specific files
>submitted via oozie command line job submissions? I'm finding that my
>Linux task controllers are breaking because they don't have permissions
>to read the appTokens .
Oozie permissions, task controllers, appTokens
Posted by Jay Vyas <ja...@gmail.com>.
What is the permissions model for appTokens and other job specific files submitted via oozie command line job submissions? I'm finding that my Linux task controllers are breaking because they don't have permissions to read the appTokens .
Re: oozie java action : Why is the nameNode required?
Posted by Robert Kanter <rk...@cloudera.com>.
You would put something equivalent instead of the NN. For example, when
running against the local filesystem, you’d do:
<name-node>file:////</name-node>
I agree that its not exactly clear.
In any case, lets keep the discussion for this on OOZIE-1695 (which you
created) to keep it all in one place.
On Thu, Feb 13, 2014 at 1:23 PM, Jay Vyas <ja...@gmail.com> wrote:
> Thanks mona:
>
> But... what if we are running oozie without HDFS (i.e. on S3Filesystem,
> GlusterFileSystem, or any other file system which has no NameNode )? This
> is clearly supported (see OOZIE-426).
>
>
> On Thu, Feb 13, 2014 at 3:04 PM, Mona Chitnis <ch...@yahoo-inc.com>
> wrote:
>
> > Oozie is coupled with Map-reduce framework and uses HDFS as one of the
> > supported filesystems. Oozie will load your workflow definition from the
> > HDFS location so NN required.
> > If you use xml namespace version 0.4 and higher, you can use a <global>
> > tag to specify common params such as NN/JT/configuration for all your
> > actions.
> >
> > ‹Mona
> >
> > On 2/13/14, 11:47 AM, "Jay Vyas" <ja...@gmail.com> wrote:
> >
> > >Is a nameNode a required element for plain java or mapreduce oozie
> > >actions?
> > >It seems so, but im not sure why:
> > >
> > > oozie job -oozie http://localhost:11000/oozie -config job.properties
> > -run
> > >Error: E0701 : E0701: XML schema error, cvc-complex-type.2.4.a: Invalid
> > >content was found starting with element 'configuration'. One of
> > >'{"uri:oozie:workflow:0.1":name-node}' is expected.
> > >
> > >It seems like the namenode should be an optional parameter, since
> > >according
> > >to https://issues.apache.org/jira/browse/OOZIE-426, OOZIE is written
> > >directly against the file system API, without any coupling to HDFS/.
> > >
> > >
> > >--
> > >Jay Vyas
> > >http://jayunit100.blogspot.com
> >
> >
>
>
> --
> Jay Vyas
> http://jayunit100.blogspot.com
>
Re: oozie java action : Why is the nameNode required?
Posted by Jay Vyas <ja...@gmail.com>.
Thanks mona:
But... what if we are running oozie without HDFS (i.e. on S3Filesystem,
GlusterFileSystem, or any other file system which has no NameNode )? This
is clearly supported (see OOZIE-426).
On Thu, Feb 13, 2014 at 3:04 PM, Mona Chitnis <ch...@yahoo-inc.com> wrote:
> Oozie is coupled with Map-reduce framework and uses HDFS as one of the
> supported filesystems. Oozie will load your workflow definition from the
> HDFS location so NN required.
> If you use xml namespace version 0.4 and higher, you can use a <global>
> tag to specify common params such as NN/JT/configuration for all your
> actions.
>
> ‹Mona
>
> On 2/13/14, 11:47 AM, "Jay Vyas" <ja...@gmail.com> wrote:
>
> >Is a nameNode a required element for plain java or mapreduce oozie
> >actions?
> >It seems so, but im not sure why:
> >
> > oozie job -oozie http://localhost:11000/oozie -config job.properties
> -run
> >Error: E0701 : E0701: XML schema error, cvc-complex-type.2.4.a: Invalid
> >content was found starting with element 'configuration'. One of
> >'{"uri:oozie:workflow:0.1":name-node}' is expected.
> >
> >It seems like the namenode should be an optional parameter, since
> >according
> >to https://issues.apache.org/jira/browse/OOZIE-426, OOZIE is written
> >directly against the file system API, without any coupling to HDFS/.
> >
> >
> >--
> >Jay Vyas
> >http://jayunit100.blogspot.com
>
>
--
Jay Vyas
http://jayunit100.blogspot.com
Re: oozie java action : Why is the nameNode required?
Posted by Mona Chitnis <ch...@yahoo-inc.com>.
Oozie is coupled with Map-reduce framework and uses HDFS as one of the
supported filesystems. Oozie will load your workflow definition from the
HDFS location so NN required.
If you use xml namespace version 0.4 and higher, you can use a <global>
tag to specify common params such as NN/JT/configuration for all your
actions.
‹Mona
On 2/13/14, 11:47 AM, "Jay Vyas" <ja...@gmail.com> wrote:
>Is a nameNode a required element for plain java or mapreduce oozie
>actions?
>It seems so, but im not sure why:
>
> oozie job -oozie http://localhost:11000/oozie -config job.properties -run
>Error: E0701 : E0701: XML schema error, cvc-complex-type.2.4.a: Invalid
>content was found starting with element 'configuration'. One of
>'{"uri:oozie:workflow:0.1":name-node}' is expected.
>
>It seems like the namenode should be an optional parameter, since
>according
>to https://issues.apache.org/jira/browse/OOZIE-426, OOZIE is written
>directly against the file system API, without any coupling to HDFS/.
>
>
>--
>Jay Vyas
>http://jayunit100.blogspot.com