Hi, I've developed a new jobcomp plugin that stores job completion data in an elasticsearch server. For more information, see the attached abstract: "hpckp_abstract.pdf"
The plugin has been tested and is currently working successfully in a real cluster production environment with 126 compute nodes and 2 login nodes. Plugin has been accepted to be presented on the next Slurm Training'15, which is going to be held on February, 5th-6th in Barcelona (Spain). The purpose of this email is to ask on how to proceed so the SLURM community may test the plugin and if accepted and useful, it would be a big satisfaction if it becomes merged for future stables releases. So, for the plugin development, I've had to create and modify these files. All paths are relative to the slurm root directory. CREATED src/plugins/jobcomp/elasticsearch/ CREATED src/plugins/jobcomp/elasticsearch/jobcomp_elasticsearch.c CREATED src/plugins/jobcomp/elasticsearch/Makefile CREATED src/plugins/jobcomp/elasticsearch/Makefile.am CREATED src/plugins/jobcomp/elasticsearch/Makefile.in CREATED src/plugins/jobcomp/elasticsearch/.deps/ CREATED src/plugins/jobcomp/elasticsearch/.deps/jobcomp_elasticsearch.Plo MODIFIED src/plugins/jobcomp/Makefile.am CREATED auxdir/x_ac_curl.m4 MODIFIED auxdir/Makefile.am MODIFIED configure.ac Notes: 1. Plugin uses libcurl-devel, so it has to be installed on the system. (Needed to handle the HTTP requests/reponses to the elasticsearch server). 2. New ./configure options have been added: --with-libcurl, --without-libcurl 3. Plugin can be enabled and configured through slurm.conf, here is an example: JobCompType=jobcomp/elasticsearch JobCompLoc=http://yourelasticserver:9200 So main questions are: - Should I create a github repository with all these files? - Should repository also include 14.11.3 rest of files? - Or is it a better way to show these files? Thanks so much, Alejandro Sanchez
hpckp_abstract.pdf
Description: Adobe PDF document
