File header is inconsistent among processes

Wei-keng Liao wkliao at eecs.northwestern.edu
Tue Sep 30 10:42:00 CDT 2014 

Hi, Sean

The warning message printed from PnetCDF indicates the global attribute
named "WRF_ALARM_SECS_TIL_NEXT_RING_52" has inconsistent values among MPI processes.
The source codes that cause the inconsistency is in line 366 of file share/output_wrf.F
where the value of this attribute is a timer (in the unit of seconds) set on each
process independently when calling WRFU_TimeIntervalGet(). The call to
wrf_put_dom_ti_integer() in line 369/371 will write the global attribute to the
netCDF file and hence PnetCDF caught the inconsistent attribute.
Let me know if this helps.

 361  IF ( i .LT. 10 ) THEN
 362    write(alarmname,'("WRF_ALARM_SECS_TIL_NEXT_RING_0",i1)')i
 363  ELSE
 364    write(alarmname,'("WRF_ALARM_SECS_TIL_NEXT_RING_",i2)')i
 365  ENDIF
 366  CALL WRFU_TimeIntervalGet(interval,S=seconds)
 367  CALL WRFU_TimeIntervalGet(tmpinterval,S=seconds2)
 368  IF ( seconds .GE. 1700000000 .OR. seconds .LE. -1700000000 ) THEN   ! it is a forever value, do not change it
 369    CALL wrf_put_dom_ti_integer( fid, TRIM(alarmname), seconds, 1, ierr )
 370  ELSE
 371    CALL wrf_put_dom_ti_integer( fid, TRIM(alarmname), seconds-seconds2, 1, ierr )
 372  ENDIF
 

Wei-keng

On Sep 30, 2014, at 9:06 AM, Sean Byland wrote:

> Thanks Wei-king,
> I’ve attached the ncdump output for one of the two failed domains. When
> the user ran wrf, pnetcdf reported a problem when writing the wrf restart
> file. When I ran it it appears to have completed the restart files but the
> file that’s incomplete is the wrf output file for the second/third domain.
> I do see a lot of these warnings in the rsl output files:
> 
> grep -r -i "inconsistent" rsl.out* | cut -d ':' -f2- | sort -u
> Warning (inconsistent metadata): attribute
> "WRF_ALARM_SECS_TIL_NEXT_RING_52" INT (676375365 != -1459748865)
> Warning (inconsistent metadata): attribute
> "WRF_ALARM_SECS_TIL_NEXT_RING_52" INT (676375365 != -1462916539)
> Warning (inconsistent metadata): attribute
> "WRF_ALARM_SECS_TIL_NEXT_RING_52" INT (676375365 != -1471304891)
> Warning (inconsistent metadata): attribute
> "WRF_ALARM_SECS_TIL_NEXT_RING_52" INT (676375365 != 1758243397)
> Warning (inconsistent metadata): attribute
> "WRF_ALARM_SECS_TIL_NEXT_RING_52" INT (676375365 != -389109179)
> Warning (inconsistent metadata): attribute
> "WRF_ALARM_SECS_TIL_NEXT_RING_52" INT (676375365 != -397497531)
> 
> Sean
> 
> 
> 
> On 9/29/14, 4:59 PM, "Wei-keng Liao" <wkliao at eecs.northwestern.edu> wrote:
> 
>> Hi, Sean
>> 
>> My understanding of WRF (wrf_io.F90, line around 1360) is when
>> NFMPI_ENDDEF() returns an error code,
>> the program returns without continuing to write any data. So, I guess the
>> file with a much smaller size
>> probably is because of this (but you should see additional error messages
>> printed on stdout.)
>> 
>> Could you show us the output of command "ncdump -h"?
>> 
>> Wei-keng
>> 
>> On Sep 29, 2014, at 4:25 PM, Sean Byland wrote:
>> 
>>> By "problem with his application” I mean unrelated (upstream) of their
>>> usage of parallel-netcdf.
>>> 
>>> Sean B.
>>> 
>>> On 9/29/14, 4:21 PM, "Sean Byland" <seanb at cray.com> wrote:
>>> 
>>>> Thanks. When I ran his application one of the output files was an
>>>> order of
>>>> magnitude too small but parallel-netcdf didn’t report a problem making
>>>> PNETCDF_SAFE_MODE less useful. This makes me think there’s a problem
>>>> with
>>>> his application (i.e. a race condition).
>>>> 
>>>> Sean B.
>>>> 
>>>> On 9/29/14, 4:05 PM, "Rob Latham" <robl at mcs.anl.gov> wrote:
>>>> 
>>>>> 
>>>>> 
>>>>> On 09/29/2014 03:56 PM, Sean Byland wrote:
>>>>>> Rob,
>>>>>> Unfortunately I wasn’t able to reproduce the error, but he did have
>>>>>> an
>>>>>> ncfile laying around so I can provide the ncdump -h output. Not to be
>>>>>> dense but what should I be looking for (I see lots of time values) ?
>>>>>> Knowing almost nothing about pnetcdf, I would think that if different
>>>>>> processes had inconsistent data, wouldn’t they fail on the write and
>>>>>> therefore I wouldn’t be able to observe what values where
>>>>>> inconsistent
>>>>>> ?
>>>>> 
>>>>> I think you're going to be better served by Wei-keng's environment
>>>>> variable suggestion, but read on for a bit of background:
>>>>> 
>>>>> Parallel-NetCDF expects the header to be identical on all N MPI
>>>>> processes.  How can processes have different data and yet still read
>>>>> the
>>>>> file?  well, the header is pretty simple.    It's not too far off to
>>>>> think of it as a big array of (now 64 bit) values.
>>>>> 
>>>>> On one process you might have the attribute "timestamp" with a value
>>>>> "2014-09-29-16:00:34 CST", followed by the information "the variable
>>>>> Pressure starts at offset 20023423 bytes".
>>>>> 
>>>>> On another process, you might have the exact same information, except
>>>>> the time stamp is "2014-09-29-23:00:34 GMT".  The information about
>>>>> the
>>>>> variable will still start at the same place and contain the same
>>>>> information.  Rank 0 will broadcast its version of the header to all
>>>>> the
>>>>> other processes.  If any of them differ in any byte, the library will
>>>>> give an error.
>>>>> 
>>>>> If the check was more involved, we could warn about attribute values
>>>>> that differ slightly but press on if "important" values (which we
>>>>> would
>>>>> have to define) were all consistent.
>>>>> 
>>>>> ==rob
>>>>> 
>>>>>> 
>>>>>> Thanks for any info.
>>>>>> 
>>>>>> Sean
>>>>>> 
>>>>>> 
>>>>>> On 9/19/14, 1:42 PM, "Rob Latham" <robl at mcs.anl.gov> wrote:
>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> On 09/19/2014 12:54 PM, Sean Byland wrote:
>>>>>>>> Hello,
>>>>>>>> A parallel-netcdf user gets this error:
>>>>>>>> 
>>>>>>>>   NetCDF error: File header is inconsistent among processes
>>>>>>>>   NetCDF error ( -250 ) from NFMPI_ENDDEF in
>>>>>>>> ext_pnc_open_for_write_commit wrf_io.F90, line 1360
>>>>>>>>  med_restart_out: opening wrfrst_d01_2013-06-01_00_10_00 for
>>>>>>>> writing
>>>>>>> 
>>>>>>> The most common cause for this message is when there's a timestamp
>>>>>>> attribute: the processes are not 100% in lock step, and so create
>>>>>>> ever
>>>>>>> so slightly different timestamps.
>>>>>>> 
>>>>>>> Can you provide the header of a CCE run? You can use 'ncmpidump -h'
>>>>>>> or
>>>>>>> serial netcdf's 'ncdump -h'
>>>>>>> 
>>>>>>> ==rob
>>>>>>> 
>>>>>>>> 
>>>>>>>> when running  software/parallel-netcdf (1.5.0) libraries that were
>>>>>>>> built
>>>>>>>> with CCE but doesn’t with an application/library that were built
>>>>>>>> with
>>>>>>>> Intel’s compiler. I’m still waiting on the user for something that
>>>>>>>> I
>>>>>>>> need to reproduce the error and start experimenting but was hoping
>>>>>>>> that
>>>>>>>> someone on this mailing list might have some useful information or
>>>>>>>> hints
>>>>>>>> about what’s causing this error and how I might fix it (or where I
>>>>>>>> might
>>>>>>>> look). All of the “make check” test pass. Thanks for any input.
>>>>>>>> 
>>>>>>>> Sean B
>>>>>>> 
>>>>>>> --
>>>>>>> Rob Latham
>>>>>>> Mathematics and Computer Science Division
>>>>>>> Argonne National Lab, IL USA
>>>>>> 
>>>>> 
>>>>> -- 
>>>>> Rob Latham
>>>>> Mathematics and Computer Science Division
>>>>> Argonne National Lab, IL USA
>>>> 
>>> 
>> 
> 
> <ncdump.wrfout_d02.out>

More information about the parallel-netcdf mailing list