[yt-users] IOError in h5py
Matthew Turk
matthewturk at gmail.com
Mon Jan 7 08:19:18 PST 2013
Hi Patrick,
Ah, interesting -- this is before the fix to the GDF frontend. A
couple things -- it looks like you've made some changes to the source
code (the + at the end of the hash indicates this.) If this isn't a
big deal, or you don't need those changes, I'd recommend you update to
the latest version manually:
cd /home/patrick/Programme/yt-x86_64/src/yt-hg
hg pull
hg up -C
python2.7 setup.py build_ext -i
If you want to retain those changes, you can be a bit safer about it:
cd /home/patrick/Programme/yt-x86_64/src/yt-hg
hg diff > backup.patch
hg pull
hg up
python2.7 setup.py build_ext -i
This will prompt you to merge your changes into the tip of the
development repository. If the merge fails, you can use the contents
of "backup.patch" to restore your changes later.
Can you try updating and see if that fixes the issue?
-Matt
On Mon, Jan 7, 2013 at 11:16 AM, Patrick Rieser
<patrick.rieser at uibk.ac.at> wrote:
> Here you go:
>
> yt module located at:
> /home/patrick/Programme/yt-x86_64/src/yt-hg
> The supplemental repositories are located at:
> /home/patrick/Programme/yt-x86_64/src/yt-supplemental
>
> The current version of the code is:
>
> ---
> 9d5b945ee428+ (yt) tip
> ---
>
> This installation CAN be automatically updated.
> Updated successfully.
>
>
> Best wishes,
> Patrick
>
> Am 2013-01-07 17:04, schrieb Matthew Turk:
>
>> Hi Patrick,
>>
>> This could also be related to a critical bug fix identified not too
>> long ago. One tricky thing about HDF5 is that once in a while
>> depending on the order of operations, errors can accumulate without
>> being flushed. What may have happened is that the file was left open
>> at a different juncture and then never closed; this could result in a
>> non-intuitive error message. What we identified was that the
>> GridDataFormat frontend was in fact leaving the files open, which then
>> left the handle around and caused non-intuitive errors later on during
>> the process. Could you send the output of:
>>
>> yt instinfo
>>
>> This will help us determine how the error is occurring.
>>
>> Best,
>>
>> Matt
>>
>> On Mon, Jan 7, 2013 at 11:02 AM, Patrick Rieser
>> <patrick.rieser at uibk.ac.at> wrote:
>>>
>>> In iyt I can load a snapshot just fine.
>>>
>>> It could be a problem using the Sun Grid Engine then. Here is my python
>>> submit script if you need it (it sources the yt environment, does some
>>> things on the data set and then submits a yt job for a subset):
>>> http://paste.yt-project.org/show/3045/
>>>
>>>
>>> h5ls output seems ok:
>>> bflags Dataset {2633, 1}
>>> block\ size Dataset {2633, 3}
>>> bounding\ box Dataset {2633, 3, 2}
>>> coordinates Dataset {2633, 3}
>>> dens Dataset {2633, 16, 16, 16}
>>> gid Dataset {2633, 15}
>>> integer\ runtime\ parameters Dataset {46}
>>> integer\ scalars Dataset {14}
>>> logical\ runtime\ parameters Dataset {53}
>>> logical\ scalars Dataset {2}
>>> magp Dataset {2633, 16, 16, 16}
>>> magx Dataset {2633, 16, 16, 16}
>>> magy Dataset {2633, 16, 16, 16}
>>> magz Dataset {2633, 16, 16, 16}
>>> node\ type Dataset {2633}
>>> pden Dataset {2633, 16, 16, 16}
>>> pres Dataset {2633, 16, 16, 16}
>>> processor\ number Dataset {2633}
>>> real\ runtime\ parameters Dataset {76}
>>> real\ scalars Dataset {15}
>>> refine\ level Dataset {2633}
>>> sim\ info Dataset {1}
>>> string\ runtime\ parameters Dataset {78}
>>> string\ scalars Dataset {1}
>>> temp Dataset {2633, 16, 16, 16}
>>> unknown\ names Dataset {11, 1}
>>> velx Dataset {2633, 16, 16, 16}
>>> vely Dataset {2633, 16, 16, 16}
>>> velz Dataset {2633, 16, 16, 16}
>>> which\ child Dataset {2633}
>>>
>>>
>>> Could the problem be caused by a defunct snapshot that got mixed in?
>>>
>>>
>>> Best wishes,
>>> Patrick
>>>
>>>
>>> Am 2013-01-07 16:54, schrieb j s oishi:
>>>
>>> Hi Patrick,
>>>
>>> What happens if you do
>>>
>>> $ h5ls <filename>
>>>
>>> where <filename> is the file you're trying to open? This looks like it
>>> could
>>> be an error outside of yt.
>>>
>>> Also, can you load() any other snapshots?
>>>
>>> Jeff
>>>
>>>
>>> On Mon, Jan 7, 2013 at 10:49 AM, Patrick Rieser
>>> <patrick.rieser at uibk.ac.at>
>>> wrote:
>>>>
>>>> I stumbled over some error in yt while executing a script that used to
>>>> work on an older version of yt. Here is the traceback:
>>>>
>>>> File "start_SGE.py", line 28, in <module>
>>>> File "/home/patrick/Auswertung/yt-run/Find_Clumps/get_name.py", line
>>>> 58,
>>>> in get_snapshots
>>>> File "/home/patrick/Auswertung/yt-run/Find_Clumps/get_name.py", line
>>>> 43,
>>>> in get_redshift
>>>> File "/home/patrick/Programme/yt-x86_64/src/yt-hg/yt/convenience.py",
>>>> line 71, in load
>>>> File
>>>>
>>>> "/home/patrick/Programme/yt-x86_64/src/yt-hg/yt/frontends/flash/data_structures.py",
>>>> line 218, in __init__
>>>> File
>>>>
>>>> "/home/patrick/Programme/yt-x86_64/lib/python2.7/site-packages/h5py/_hl/files.py",
>>>> line 165, in __init__
>>>> File
>>>>
>>>> "/home/patrick/Programme/yt-x86_64/lib/python2.7/site-packages/h5py/_hl/files.py",
>>>> line 57, in make_fid
>>>> File "h5f.pyx", line 70, in h5py.h5f.open (h5py/h5f.c:1626)
>>>> IOError: unable to open file (File accessability: Unable to open file)
>>>> Exception AttributeError: "'NoneType' object has no attribute 'close'"
>>>> in
>>>> ignored
>>>>
>>>>
>>>> I'd be very thankful if somebody got any idea how to fix this!
>>>>
>>>> Best wishes,
>>>> Patrick
>>>>
>>>> _______________________________________________
>>>> yt-users mailing list
>>>> yt-users at lists.spacepope.org
>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> yt-users mailing list
>>> yt-users at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>
>>>
>>>
>>> _______________________________________________
>>> yt-users mailing list
>>> yt-users at lists.spacepope.org
>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>>
>> _______________________________________________
>> yt-users mailing list
>> yt-users at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>
>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
More information about the yt-users
mailing list