[CIG-SHORT] Question About PyLith

[Brad Aagaard]

Wang Guiwen,

The examples are small enough that they do not need to be run in 
parallel. If you did run any of them in parallel, then you probably 
didn't use more than one compute node.

To run on a cluster in parallel on more than one compute node, you 
*must* build from source. We strongly recommend using the PyLith 
installer. If this is difficult, then we recommend getting help from 
whoever provides user support for your cluster, such as a system 
administrator.

The log you sent does show an error:

RuntimeError: Error querying for 'change' at (  0  18202.9  -600.055) 
using spatial database 'Traction perturbation'. I believe you are 
getting this error because the parameter file has not been updated to 
include the marking of the buried edges. See the PyLith manual and 
tutorials for how to do this.

Regards,
Brad

On 12/9/18 5:25 AM, wanggw at sxnu.edu.cn wrote:
> Hi sir:
> Thank you for your reply; I use binary on my cluster,because compiling 
> from source is too complex .I have to give up.It seems that i must 
> compile the source.
> The Pylith examples in the manual can  run correctly in my cluster ,so I 
> try to run the tpv102.
>          I capture both stdout and stderr to the file hex8_200m.log.
> 
>        Best wishes to you!
> 
>                  Wang Guiwen
>                                                                          
>                  2018.12.09
> 
> ------------------------------------------------------------------------
> shanxi normal university China
> wanggw at sxnu.edu.cn
> 
>     *From:* Brad Aagaard <mailto:baagaard at usgs.gov>
>     *Date:* 2018-12-08 03:48
>     *To:* wanggw <mailto:wanggw at sxnu.edu.cn>
>     *CC:* cig-short at geodynamics.org <mailto:cig-short at geodynamics.org>
>     *Subject:* Re: Question About PyLith
>     Wang Guiwen,
>     I can't tell from the output (you need to capture both stdout and
>     stderr
>     to the log, for example >& hex8_200m.log).
>     Some possible issues:
>     1. You need to build from source if you are running on a cluster. Use
>     the pylith_installer. The binary will not work on multiple compute
>     nodes
>     of a cluster, because it uses MPI that is limited to a desktop or a
>     single compute node.
>     2. The benchmark repository includes benchmarks that were created over
>     10 years and many have not been updated for the current version. See
>     the
>     PyLith release notes for changes needed to update old parameter files.
>     We would greatly appreciate users submitting pull requests that update
>     the parameter files.
>     Regards,
>     Brad
> 
>