[Swift-devel] Error messages and execution retries

[Michael Wilde]

I think your note just before this, Ben, confirms my suspicion.

The improvement needed, I believe, is the more immediate and unbuffered posting of useful error info in a way that makes it less likely to be lost in the event swift never reaches its normal exit processing.

- Mike

----- "Michael Wilde" <wilde at mcs.anl.gov> wrote:

> We should probably test to make sure that this is the case. For
> example, if 10 jobs are launched in parallel, and one fails, then even
> with lazy.errors=false, the 9 running jobs will still finish, right?
> Its just that no new ones will start. So will the error (stderr?) from
> the failing job be sent to the log or the swift stdout/err right away,
> or will it still wait for the full swift termination, which still may
> get circumvented by a ^C ???
> 
> I'm rather suspicious that its the latter (undesirable) case, although
> I may be wrong.
> 
> - Mike
> 
> 
> ----- "Mihael Hategan" <hategan at mcs.anl.gov> wrote:
> 
> > That's only if lazy errors are enabled. Perhaps they should be
> > disabled
> > by default.
> > 
> > On Wed, 2010-08-18 at 14:23 -0600, Michael Wilde wrote:
> > > I think the case is that these errors are only presented when the
> > swift command reaches "normal" compete termination, which in early
> > debugging of a script is a rare event. By "normal" I mean all
> runnable
> > tasks are done, at which time more complete messages for prior
> errors
> > are emitted.
> > > 
> > > - Mike
> > > 
> > > 
> > > ----- "Ben Clifford" <benc at hawaga.org.uk> wrote:
> > > 
> > > > even when all retries have happened?
> > > > 
> > > > -- 
> > > > http://www.hawaga.org.uk/ben/
> > > > 
> > > > _______________________________________________
> > > > Swift-devel mailing list
> > > > Swift-devel at ci.uchicago.edu
> > > > http://mail.ci.uchicago.edu/mailman/listinfo/swift-devel
> > >
> 
> -- 
> Michael Wilde
> Computation Institute, University of Chicago
> Mathematics and Computer Science Division
> Argonne National Laboratory
> 
> _______________________________________________
> Swift-devel mailing list
> Swift-devel at ci.uchicago.edu
> http://mail.ci.uchicago.edu/mailman/listinfo/swift-devel

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