Could I modify /misc/local/galaxy/galaxy-dist/lib/galaxy/datatypes/sniff.py to print out debug information like host, os.environ, tempfile.gettempdir(), etc?
Would I be able to see its stdout from galaxy or the log, or is there something special I need to do to retrieve the information? On Thu, Feb 7, 2013 at 1:12 PM, greg <margeem...@gmail.com> wrote: > Update: > > When I run as the Galaxy user, Python does have the right temp directory: > >>>> tempfile.gettempdir() > '/scratch/galaxy' > > So does that mean this upload job isn't running as galaxy, or is > skipping the job_environment_setup_file? Or could something else be > going on? > > Any ideas, now I'm really stuck. > > Thanks, > > Greg > > > > On Wed, Feb 6, 2013 at 3:35 PM, greg <margeem...@gmail.com> wrote: >> Ok, when I ran Python in my last two emails I was running as myself, >> not the galaxy user, and only the galaxy user has write permission to >> /scratch/galaxy >> >> So that's why Python was ignoring /scratch/galaxy for me. If it >> doesn't have write access it tries the next temp directory in its >> list. >> >> I'm going to try debugging as the galaxy user next. >> >> -Greg >> >> On Wed, Feb 6, 2013 at 3:21 PM, greg <margeem...@gmail.com> wrote: >>> Hi Nate, >>> >>> I don't see $TMPDIR being set on the cluster, in addition to my >>> previous email I ran: >>> >>>>>> print os.environ.keys() >>> ['KDE_IS_PRELINKED', 'FACTERLIB', 'LESSOPEN', 'SGE_CELL', 'LOGNAME', >>> 'USER', 'INPUTRC', 'QTDIR', 'PATH', 'PS1', 'LANG', 'KDEDIR', 'TERM', >>> 'SHELL', 'TEMP', 'QTINC', 'G_BROKEN_FILENAMES', 'SGE_EXECD_PORT', >>> 'HISTSIZE', 'KDE_NO_IPV6', 'MANPATH', 'HOME', 'SGE_ROOT', 'QTLIB', >>> 'VIRTUAL_ENV', 'SGE_CLUSTER_NAME', '_', 'SSH_CONNECTION', 'SSH_TTY', >>> 'HOSTNAME', 'SSH_CLIENT', 'SHLVL', 'PWD', 'MAIL', 'LS_COLORS', >>> 'SGE_QMASTER_PORT'] >>> >>> But I think we've narrowed it down to something interfering with >>> Python deciding the temp file location. I just can't figure out what. >>> >>> >>> >>> On Wed, Feb 6, 2013 at 3:18 PM, Nate Coraor <n...@bx.psu.edu> wrote: >>>> On Feb 6, 2013, at 3:00 PM, greg wrote: >>>> >>>>> Thanks Nate, >>>>> >>>>> It turns out I already had this as the first line of my job setup file: >>>>> >>>>> export TEMP=/scratch/galaxy >>>>> >>>>> But when I look in that directory, there's plenty of free space, and I >>>>> also don't see any recent files there. So I'm wondering if the upload >>>>> jobs aren't seeing that for some reason. >>>>> >>>>> Any ideas on how I could diagnose this more? >>>> >>>> Hi Greg, >>>> >>>> The first place to look would be in lib/galaxy/datatypes/sniff.py, line 96: >>>> >>>> fd, temp_name = tempfile.mkstemp() >>>> >>>> If you print temp_name, that will tell you what file the upload tool is >>>> writing to. You may also want to take a look at: >>>> >>>> http://docs.python.org/2/library/tempfile.html#tempfile.tempdir >>>> >>>> Some cluster environments set $TMPDIR, and if that is set, $TEMP will not >>>> be used. >>>> >>>> --nate >>>> >>>>> >>>>> -Greg >>>>> >>>>> >>>>> Relevant info? >>>>> >>>>> >>>>>> grep env galaxy-dist/universe_wsgi.ini >>>>> environment_setup_file = /usr/local/galaxy/job_environment_setup_file >>>>> >>>>>> cat /usr/local/galaxy/job_environment_setup_file >>>>> export TEMP=/scratch/galaxy >>>>> #active Python virtual env just for galaxy >>>>> source /usr/local/galaxy/galaxy_python/bin/activate >>>>> ... path setup lines ... >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> On Wed, Feb 6, 2013 at 2:37 PM, Nate Coraor <n...@bx.psu.edu> wrote: >>>>>> On Feb 6, 2013, at 2:32 PM, greg wrote: >>>>>> >>>>>>> Hi guys, >>>>>>> >>>>>>> When I try to upload a directory of files from a server directory I'm >>>>>>> seeing the error below. >>>>>>> >>>>>>> It appears to be trying to write to a temp directory somewhere that >>>>>>> I'm guessing doesn't have enough space? Is there a way I can direct >>>>>>> where it writes to for temporary files like this? >>>>>> >>>>>> Hi Greg, >>>>>> >>>>>> There are a few ways. For some parts of Galaxy, you will want to set >>>>>> new_file_path in universe_wsgi.ini to a suitable temp space. However, >>>>>> this is not the case for the upload tool. >>>>>> >>>>>>> Am I understanding right, that these upload jobs are running on our >>>>>>> cluster? I think it would be a problem if its trying to use the >>>>>>> default temp directory on each cluster node since they aren't >>>>>>> provisioned with much space. >>>>>> >>>>>> This is correct. On the cluster, add something to your user's shell >>>>>> startup files (or see the environment_setup_file option in >>>>>> universe_wsgi.ini) that will set the $TEMP or $TMPDIR environment >>>>>> variable to a suitable temp space. >>>>>> >>>>>> --nate >>>>>> >>>>>>> >>>>>>> Please advise. >>>>>>> >>>>>>> Thanks, >>>>>>> >>>>>>> Greg >>>>>>> >>>>>>> >>>>>>> >>>>>>> Miscellaneous information:Traceback (most recent call last): File >>>>>>> "/misc/local/galaxy/galaxy-dist/tools/data_source/upload.py", line >>>>>>> 384, in __main__() File >>>>>>> "/misc/local/galaxy/galaxy-dist/tools/data_source/upload.py", line >>>>>>> 373, in __main__ add_file( dataset, >>>>>>> Job Standard Error >>>>>>> >>>>>>> >>>>>>> Traceback (most recent call last): >>>>>>> File "/misc/local/galaxy/galaxy-dist/tools/data_source/upload.py", >>>>>>> line 384, in >>>>>>> __main__() >>>>>>> File "/misc/local/galaxy/galaxy-dist/tools/data_source/upload.py", >>>>>>> line 373, in __main__ >>>>>>> add_file( dataset, registry, json_file, output_path ) >>>>>>> File "/misc/local/galaxy/galaxy-dist/tools/data_source/upload.py", >>>>>>> line 270, in add_file >>>>>>> line_count, converted_path = sniff.convert_newlines( dataset.path, >>>>>>> in_place=in_place ) >>>>>>> File "/misc/local/galaxy/galaxy-dist/lib/galaxy/datatypes/sniff.py", >>>>>>> line 99, in convert_newlines >>>>>>> fp.write( "%s\n" % line.rstrip( "\r\n" ) ) >>>>>>> IOError: [Errno 28] No space left on device >>>>>>> ___________________________________________________________ >>>>>>> Please keep all replies on the list by using "reply all" >>>>>>> in your mail client. To manage your subscriptions to this >>>>>>> and other Galaxy lists, please use the interface at: >>>>>>> >>>>>>> http://lists.bx.psu.edu/ >>>>>> >>>> ___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
