New development workflow of the Configuration Management

Thursday 26 Sept 2013, 14:00 → 15:00 Europe/Zurich

CERN

Slides VOCsMeeting20130926.pdf VOCsMeeting20130926.pptx

Attendance

IT-CIS       Marek Domaracky
IT-CS        Véronique Lefebure
IT-DB        Georgios Kaklamanos
IT-DI         Denise Heagerty
IT-DSS      Alex Iribarren, Jan Iven
IT-OIS       Belmiro Mareira, Jan van Eldik
IT-PES       Alex Lossent, Maite Barroso, Ignacio Reguero, Nacho Barrientos, Ioannis Agtzidis, Vítor Gouveia
ATLAS  Alexey BuzyKaev, Sergey Baranov, Yuri Smirnov
LCD      Andre Sailer, Christian Grefe
CMS     Ivan Glushkov, Diego Da Silva Gomes

Questions

JI: what happens to machines after the hostgroup cleanup end September?

• If any still exist in the hostgroups, virtual or physical machines themselves will not be touched, but not registered in Foreman anymore (so no Puppet configuration anymore)

CMS: Do we need to create a FE to define the ownership of a module? What about PH, which is not using Service-Now?

• An e-group can be specified in place of the FE. This is also an acceptable method to contact the MM.

AL: how do MM get notified about problems that users report about a shared module?

• This is not finalized. Fire a ticket in Jira and pray.

JI: who has powers to modify existing environment definitions?

• Everybody is able to modify all environment definitions

CMS: pull requests available?

• Not in central git service

ATLAS: if no special environment file, what is the default?

• Qa or master are available to all machines (and will use the corresponding branch for each and every module). It's specified at machine creation and can be modified in Foreman.

JI: do I have to monitor my machines in QA for any failure in order to detect problems? E.g. an alarm in case of Puppet run failure?

• Yes, but we're looking at ways to automate the process/automate tests.
• It will also be possible to freeze the configuration for some time (e.g. while you're on holidays – but you can also stop the Puppet service)

IG: How to get a Puppet-managed physical machine or reisntall an existing Quattor node with Puppet? There is no documentation in the IT Configuration Management doc.

• The process for being allocated hardware for a Puppet-managed physical machine is going to be formalized and documentation will be added to the IT website. Follow the progress at https://its.cern.ch/jira/browse/AI-3133

 

14:00 → 15:00 New Development workflow of the Configuration Management