[Swift-user] Swift error with no explanation

[Michael Wilde]

Jonathan, it seems that there's some kind of error cascade here. You have an application script that repeatedly runs Swift. Several Swift runs complete successfully. Then one run encounters a Java null pointer exception in Swift, seemingly in a method related to logging progress of the run. Close to the same time, one of your app() calls is failing.  After that, all the Swift runs fail due to an app() call thats failing in a different manner.

Im moving this discussion to swift-support to avoid clogging up swift-user with lots of debugging discussion.  Please send us (on swift-support) a pointer to where we can locate your run directories.  I'd like to find the log file for RunID: 20120914-0423-eag9170a, and send that to Mihael to debug the null pointer exception.

I'd like to look closer at the subsequent runs to see whats failing. Im thinking something went wrong in your application environment thats causing the initial app failure, which in turn may have tripped into some intermittent Swift bug.

- Mike


----- Original Message -----
> From: "Jonathan Margoliash" <jmargolpeople at gmail.com>
> To: swift-user at ci.uchicago.edu, "Swift Language" <davidk at ci.uchicago.edu>, "Michael Wilde" <wilde at mcs.anl.gov>,
> "Professor E. Yan" <eyan at anl.gov>
> Sent: Friday, September 14, 2012 9:25:41 AM
> Subject: Swift error with no explanation
> Hello swift support,
> 
> 
> Last night I was running a swift script, and it crashed for no reason
> I can discern. The attached file terminal_output.txt is the output
> (both stdout and stderr) that was printed during the run. To the best
> of my knowledge, the first error crops up on line 18192. The text is
> "Execution failed: java stack trace with NullPointerException at the
> top". What's more confusing is that swift printed the following to the
> terminal:
> 
> 
> RunID: 20120914-0423-eag9170a
> Failed to transfer wrapper log for job Nelder_Mead-mnsj53yk
> EXCEPTION Exception in Nelder_Mead:
> Arguments: [SCE_Par.mat, simplex7_3.mat, new_point7_3_1.mat, 1, 151,
> run_swat.m]
> Host: thwomp
> Directory:
> generate_offspring-20120914-0423-eag9170a/jobs/m/Nelder_Mead-mnsj53yk
> stderr.txt:
> 
> 
> stdout.txt:
> 
> 
> ----
> 
> 
> The app that I was running (Nelder_Mead.sh) prints a line to the
> stdout as the first thing it does, and since swift registered nothing
> being written to stdout.txt, my only conclusion is that my app was
> never run, and that this wasn't a problem with my code. I've attached
> two log files to this email, in hopes that one of them corresponds to
> this swift run, but neither of them have the right starting time (too
> soon or too late), and my guess is that the line "Failed to transfer
> wrapper log for job Nelder_Mead-mnsj53yk" indicates that no logfile
> was generated for this call to swift. Also, please ignore the output
> to terminal_output.txt generated after the swift-crash. My program
> kept running and making more calls to swift, but I cannot predict what
> it was doing given that I don't know what happened with swift, and I
> would have rather it had just crashed when swift did.
> 
> 
> Anyway, thanks for the help,
> 
> 
> Jonathan

-- 
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory