You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Mahmood Naderan <nt...@yahoo.com> on 2014/03/14 10:47:10 UTC
debug mode
Hello,
1- Can we debug Hadoop/Mahout with the available bin files? Or we have to do rebuild them? Probably they are compiled with optimizations
2- I cannot find a procedure on "how to build Hadoop/Mahout in debug mode". If anyone knows, please let me know.
Since I am a gcc/make guy, maybe these are noob questions in Hadoop/Mahout. But I don't know the details! Thanks for any help
Regards,
Mahmood
Re: debug mode
Posted by Mahmood Naderan <nt...@yahoo.com>.
Excuse me, I forgot to say that should I use mvndebug for debugging purpose?
Regards,
Mahmood
On , Mahmood Naderan <nt...@yahoo.com> wrote:
Let me state in this way. Using GNU Make, we use "-g -ggdb" to insert debug symbols in the object file. On the other hand, if we use "-O3" it will optimize the code and remove debug symbols. As a result, further debugging is not possible.
Now I want to know, if I run "mvn install" will it optimize the code for performance by removing the debug symbols?
Regards,
Mahmood
On Thursday, March 20, 2014 4:22 AM, Pat Ferrel <pa...@occamsmachete.com> wrote:
If you are using a debugger like IntelliJ or Eclipse you just create a project that uses Mahout. By default it will run any hadoop on the native local file system with all processes on your debug machine. That is as far as I’ve needed to go.
Andrew is talking about how to debug while running the code on a cluster. This often not needed and certainly not required by Mahout.
On Mar 16, 2014, at 6:48 PM, Andrew Musselman <an...@gmail.com> wrote:
Yes, there are ways to debug.
One is to attach a remote debugger, set breakpoints, etc., like so:
https://www.google.com/search?q=attach+remote+debugger+java+example+eclipse+or+intellij
The other would be to write a log4j.properties file for Mahout and/or
Hadoop and set the logging level to "more wordy", like DEBUG. I did this
with Hadoop years ago and recently tried with Mahout but didn't get it
working.
The remote debugger would let you step through code while the DEBUG level
would open up some more informative stack traces as I understand.
On Fri, Mar 14, 2014 at 2:47 AM, Mahmood Naderan <nt...@yahoo.com>wrote:
> Hello,
> 1- Can we debug Hadoop/Mahout with the available bin files? Or we have to
> do rebuild them? Probably they are compiled with optimizations
> 2- I cannot find a procedure on "how to build Hadoop/Mahout in debug
> mode". If anyone knows, please let me know.
>
> Since I am a gcc/make guy, maybe these are noob questions in
> Hadoop/Mahout. But I don't know the details! Thanks for any help
>
>
> Regards,
> Mahmood
Re: debug mode
Posted by Mahmood Naderan <nt...@yahoo.com>.
Let me state in this way. Using GNU Make, we use "-g -ggdb" to insert debug symbols in the object file. On the other hand, if we use "-O3" it will optimize the code and remove debug symbols. As a result, further debugging is not possible.
Now I want to know, if I run "mvn install" will it optimize the code for performance by removing the debug symbols?
Regards,
Mahmood
On Thursday, March 20, 2014 4:22 AM, Pat Ferrel <pa...@occamsmachete.com> wrote:
If you are using a debugger like IntelliJ or Eclipse you just create a project that uses Mahout. By default it will run any hadoop on the native local file system with all processes on your debug machine. That is as far as I’ve needed to go.
Andrew is talking about how to debug while running the code on a cluster. This often not needed and certainly not required by Mahout.
On Mar 16, 2014, at 6:48 PM, Andrew Musselman <an...@gmail.com> wrote:
Yes, there are ways to debug.
One is to attach a remote debugger, set breakpoints, etc., like so:
https://www.google.com/search?q=attach+remote+debugger+java+example+eclipse+or+intellij
The other would be to write a log4j.properties file for Mahout and/or
Hadoop and set the logging level to "more wordy", like DEBUG. I did this
with Hadoop years ago and recently tried with Mahout but didn't get it
working.
The remote debugger would let you step through code while the DEBUG level
would open up some more informative stack traces as I understand.
On Fri, Mar 14, 2014 at 2:47 AM, Mahmood Naderan <nt...@yahoo.com>wrote:
> Hello,
> 1- Can we debug Hadoop/Mahout with the available bin files? Or we have to
> do rebuild them? Probably they are compiled with optimizations
> 2- I cannot find a procedure on "how to build Hadoop/Mahout in debug
> mode". If anyone knows, please let me know.
>
> Since I am a gcc/make guy, maybe these are noob questions in
> Hadoop/Mahout. But I don't know the details! Thanks for any help
>
>
> Regards,
> Mahmood
Re: debug mode
Posted by Pat Ferrel <pa...@occamsmachete.com>.
If you are using a debugger like IntelliJ or Eclipse you just create a project that uses Mahout. By default it will run any hadoop on the native local file system with all processes on your debug machine. That is as far as I’ve needed to go.
Andrew is talking about how to debug while running the code on a cluster. This often not needed and certainly not required by Mahout.
On Mar 16, 2014, at 6:48 PM, Andrew Musselman <an...@gmail.com> wrote:
Yes, there are ways to debug.
One is to attach a remote debugger, set breakpoints, etc., like so:
https://www.google.com/search?q=attach+remote+debugger+java+example+eclipse+or+intellij
The other would be to write a log4j.properties file for Mahout and/or
Hadoop and set the logging level to "more wordy", like DEBUG. I did this
with Hadoop years ago and recently tried with Mahout but didn't get it
working.
The remote debugger would let you step through code while the DEBUG level
would open up some more informative stack traces as I understand.
On Fri, Mar 14, 2014 at 2:47 AM, Mahmood Naderan <nt...@yahoo.com>wrote:
> Hello,
> 1- Can we debug Hadoop/Mahout with the available bin files? Or we have to
> do rebuild them? Probably they are compiled with optimizations
> 2- I cannot find a procedure on "how to build Hadoop/Mahout in debug
> mode". If anyone knows, please let me know.
>
> Since I am a gcc/make guy, maybe these are noob questions in
> Hadoop/Mahout. But I don't know the details! Thanks for any help
>
>
> Regards,
> Mahmood
Re: debug mode
Posted by Andrew Musselman <an...@gmail.com>.
Yes, there are ways to debug.
One is to attach a remote debugger, set breakpoints, etc., like so:
https://www.google.com/search?q=attach+remote+debugger+java+example+eclipse+or+intellij
The other would be to write a log4j.properties file for Mahout and/or
Hadoop and set the logging level to "more wordy", like DEBUG. I did this
with Hadoop years ago and recently tried with Mahout but didn't get it
working.
The remote debugger would let you step through code while the DEBUG level
would open up some more informative stack traces as I understand.
On Fri, Mar 14, 2014 at 2:47 AM, Mahmood Naderan <nt...@yahoo.com>wrote:
> Hello,
> 1- Can we debug Hadoop/Mahout with the available bin files? Or we have to
> do rebuild them? Probably they are compiled with optimizations
> 2- I cannot find a procedure on "how to build Hadoop/Mahout in debug
> mode". If anyone knows, please let me know.
>
> Since I am a gcc/make guy, maybe these are noob questions in
> Hadoop/Mahout. But I don't know the details! Thanks for any help
>
>
> Regards,
> Mahmood