PanDA Packaging Corner

Mandate

• Refactoring code: work on making the PanDA code generic, where different experiments can plugin their customized libraries (e.g. for data handling) and also be compatible to different database backends (currently MySQL and Oracle). We have to avoid the creation of branches and forks that we will have to support separately and that we will never be able to merge again.
• Deployment/validation: installation and configuration of the components should be simple and automated as much as possible. Steps for validation of the installation should be available.
• CI/Testing: We should target continuous integration and testing, in order to validate our code changes.
• Documentation: We need to spread the knowledge across teams and provide the necessary documentation to promote independent work.

Current stories

PanDA@EC2 (Owner: Fernando, Sergey Padolski)

• The server installation is frozen to a rather old version and should be upgraded
• New PanDa monitor has been reported to have Oracle specific queries that are incompatible towards MySQL
• Setup a parallel PanDA monitor instance (same machine, different port) to test changes for MySQL compatibility

PanDA@NRC KI (Owner: Ruslan)

PanDA for AMS - migration to MariaDB (Owner: KuoHao, Fernando)

self.execute("SET @@SESSION.DATETIME_FORMAT='%%Y/%%m/%%d %%H:%%i:%%s’")
ERROR 1238 (HY000): Variable 'datetime_format' is a read only variable

     def next(self):
        if self.backend == 'mysql’:
            return self.cur.fetchone()
        else:
            return self.cur.next()

• panda-cacheschedconfig is not compatible to MySQL

Monitor refactoring (Owner: Misha, Sergey Padolski)

• Repackaging has been done. Now there are only 2 packages (CORE for bigpandamon and ATLAS for prodsys2). ATLAS and CORE have been merged. ATLAS inherits from CORE.
• Pending is to have only one django application, e.g. same authentication mechanism for both packages

Use PanDA for COMPASS (Owner: Artem)

• Users will be Sergey Gerasimov and Elena
• Final purpose is to run jobs on lxbatch and store the output in EOS
• Artem has set up PanDA server
  • commented out code in PanDA server to bypass them
  • old version of monitor, since the newest version has some Oracle specific parts

PanDA server refactoring (Owner: Fernando, Ruslan, Ivan)

PanDA@OSG (Owner: Fernando)

• OSG(Carl, Brian) wants to host a PanDA instance for other sciences
• PanDA server deployment was standardized thanks to Carl's changes. Now need to give clear instructions on how to install, configure and validate.

PanDA server configuration (Owner: Artem)

PanDA-AliEN integration (Owner: Andrey Kondratyev)

Stories to be initiated

JEDI and other components for other experiments

Sexy component documentation