Britton,

Thank you very much, this seems to work, although not without a quirk -
there may be a bug in the yt_astro_analysis installer.

I am using the version of yt that you so kindly created for me a while
ago - the one that fixes the HOP parallel scaling bug; it also includes
some of my modifications to the artio frontend. When I installed
yt_astro_analysis per online instructions, the installer kicked out my
version of yt and activated another one (not sure if it was the system
wide one or it installed its own). I had to juggle with installing and
uninstalling yt and yt_astro_analysis until I figured out the order that
worked: I installed yt_astro_analysis first and then reinstalled my
custom yt on top of it. Then the rockstar error disappeared.

n



On 1/25/19 1:14 PM, Britton Smith wrote:
> Hi Nick,
>
> I think you're running into a bug related to running rockstar with
> Python3. This has been fixed as of the last major yt release and the
> release of the yt_astro_analysis package. I just tried a small rockstar
> test on Blue Waters using yt_astro_analysis and it worked for me. If you
> want to go that route, I recommend following the install instructions here:
> https://yt-astro-analysis.readthedocs.io/en/latest/
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__yt-2Dastro-2Danalysis.readthedocs.io_en_latest_&d=DwMFaQ&c=gRgGjJ3BkIsb5y6s49QqsA&r=7jSLZBQB_2gqGQ3elA922w&m=r4Gndd7QTUYD3ucQildHI3Q5sxVOHHymQblz3tz-oms&s=rVlwc2zGxag2GN2YXVXgI2eZzFOXuZFxTgsX1S_UUDs&e=>
> and using the --user flag when you install with pip.
>
> Britton
>
> On Wed, Jan 23, 2019 at 8:50 AM Nick Gnedin <ngnedin@gmail.com
> <mailto:ngnedin@gmail.com>> wrote:
>
>
>     Folks,
>
>     I am trying to run Rickstar on Blue Waters, and I am getting an error
>     even for a very simple setup: 1 writer and 1 reader running on a single
>     node on a small test run. The same job runs just fine on a workstation.
>
>     The script is simple:
>     ---------------
>     import yt, yt.funcs
>     import glob, sys, os.path
>
>     yt.enable_parallelism()
>     from yt.analysis_modules.halo_finding.rockstar.api import
>     RockstarHaloFinder
>
>
>     root = "../B20"
>     n = 1
>
>     dirs = ( "D" )
>
>     for dir in dirs:
>           sets = glob.glob(root+"/"+dir+"/rei20c1_a0.*/*.art")
>           sets.sort()
>           ts = yt.DatasetSeries(sets)
>           rh = RockstarHaloFinder(ts, num_readers=n, num_writers=n,
>     outbase=root+"/"+dir+"/rs", dm_only=True, particle_type='N-BODY')
>           rh.run()
>     ---------------
>
>     and the batch script is trivial:
>     --------
>     #!/bin/csh
>     #PBS -l nodes=1:ppn=32:xe
>     #PBS -l walltime=30:00,vmem=64gb
>     #PBS -q debug
>     #PBS -N rshfc.1
>     #PBS -e $PBS_JOBID.err
>     #PBS -o $PBS_JOBID.out
>     #PBS -m be
>
>     cd $PBS_O_WORKDIR
>
>     module load bwpy
>     module load bwpy-mpi
>
>     setenv APRUN_XFER_LIMITS 1
>     limit stacksize    unlimited
>
>     aprun -n 3 -N 3 python3 rs-hfc.py >& std.rshfc1
>     --------
>
>     The error I get is not very informative:
>
>     [Error] Couldn't open ^A:0!  (Err: Name or service not known)
>
>     and the ^A string varies in separate runs, c.f.
>
>     [Error] Couldn't open 344ei20:0!  (Err: Name or service not known)
>
>     which may happen if a null or corrupted pointer is printed with a %s
>     specification, but that's all I can guess. Is there a way to obtain
>     more
>     information and localize the offending code segment?
>
>
>
>
>
>
>
>
>     _______________________________________________
>     yt-users mailing list -- yt-users@python.org
>     <mailto:yt-users@python.org>
>     To unsubscribe send an email to yt-users-leave@python.org
>     <mailto:yt-users-leave@python.org>
>