[Swift-devel] hang checker fun

[Mihael Hategan]

I don't believe you. There is no SgtDim data type in that script.

Mihael

On Sun, 2011-03-27 at 16:55 -0500, Allan Espinosa wrote:
> Here it is.  The get_app() calls are simple wrappers to readData()
> 
> type offset {
>   int off;
>   int size;
> }
> type offset_file;
> (offset _off[]) mkoffset(int _size, int _group_size) {
>    offset_file file <single_file_mapper; file=@strcat("LGU/offset-",_size)>;
>    file = mkoffset_file(_size, _group_size);
>    _off = readData(file);
> }
> app (offset_file _off) mkoffset_file(int _size, int _group_size) {
>   mkoffset _size _group_size;
> }
> 
> /* TODO: data management zip jobs */
> 
> /* Main program */
> int run_id = 664;
> int agg_size = 80;
> int loc_size = 20;
> string datadir =
>   "gsiftp://gridftp.pads.ci.uchicago.edu//gpfs/pads/swift/aespinosa/science/cybershake/Results";
> 
> Station site = get_site(run_id);
> 
> Sgt sgt_var <ext; exec="getsgtvar.rb", r=run_id, s=site.name,
>   l="gsiftp://gridftp.pads.ci.uchicago.edu//gpfs/pads/swift/aespinosa/science/cybershake/SgtFiles">;
> Rupture rups[] = get_ruptures(run_id, site);
> 
> foreach rup in rups {
>   string loc_sub = @strcat(datadir, "/", site.name, "/", rup.source,
> "/", rup.index);
>   Sgt sub <ext; exec="getsub.rb", l=loc_sub, n=site.name, s=rup.source,
>       r=rup.index>;
>   string var_str[] = get_variations( site, rup,
>       "gsiftp://gridftp.pads.ci.uchicago.edu//gpfs/pads/swift/aespinosa/science/cybershake/RuptureVariations"
> );
>   Variation vars[] <array_mapper; files=var_str>;
> 
>   sub = extract(sgt_var,  site, vars[rup.size-1]);
> 
>   string seis_str[];
>   string peak_str[];
> 
>   foreach var,i in vars {
>     seis_str[i] = @strcat(loc_sub, "/Seismogram_", site.name, "_", rup.source,
>                           "_", rup.index, "_", i, ".grm");
>     peak_str[i] = @strcat(loc_sub, "/PeakVals_", site.name, "_", rup.source,
>                           "_", rup.index, "_", i, ".bsa");
>   }
> 
>   Seismogram seis[] <array_mapper; files=seis_str>;
>   PeakValue  peak[] <array_mapper; files=peak_str>;
> 
>   if(rup.size <= loc_size) {
>     /*
>      * Not worth to transfer the data. Execute on TeraGrid instead.
>      * Also execute on localhost.
>      */
>     foreach var,i in vars {
>       (seis[i], peak[i]) = seispeak_local(sub, var, site);
>     }
>   } else {if(rup.size <= agg_size) {
>     /* Execute on a single resource */
>     (seis, peak) = seispeak_agg(sub, vars, site, rup.size);
>   } else {
>     /*offset offs[] = mkoffset(rup.size, agg_size);*/
>     /*for i in offs {*/
>       /*(seis, peak) = seispeak_agg(sub, vars[i.off:i.off+off.size],*/
>                                   /*off.size);*/
>     /*}*/
>   }}
> }
> 
> 
> 2011/3/27 Mihael Hategan <hategan at mcs.anl.gov>:
> > May I see the script?
> >
> > On Fri, 2011-03-25 at 19:42 -0500, Allan Espinosa wrote:
> >> this has been occurring for 70 times already.  What i expect is for
> >> the app with SgtDim sub to run and close the future.
> >>
> >> 2011-03-25 19:40:12,217-0500 WARN  HangChecker No events in 10s.
> >> 2011-03-25 19:40:12,217-0500 WARN  HangChecker
> >> Registered futures:
> >> Rupture[] rups  Closed, 1 elements, 0 listeners
> >> Variation vars - Closed, no listeners
> >> SgtDim sub - Open, 1 listeners
> >> string site  Closed, no listeners
> >> Variation[] vars  Closed, 72 elements, 0 listeners
> >> ----
> >>
> >> Waiting threads:
> >> 0-13
> >> 0-13-0-7
> >> 0-13-0-8-1-1
> >> ----
> >>
> >>
> >
> >
> >
> >
> 
> 
>