Creating Load-Balanced Services on top of Cloud Infrastructure and Puppet

Thursday 10 Oct 2013, 14:00 → 16:00 Europe/Zurich

513/1-024 (CERN)

Slides VOCsMeeting20131010.pdf VOCsMeeting20131010.pptx

Attendance

IT-CF        Michal Wasilewski
IT-CS        Véronique Lefebure
IT-DI         Denise Heagerty
IT-DSS      Jan Iven
IT-OIS       Tim Bell,
IT-PES       Alex Lossent, Ignacio Reguero, Nacho Barrientos, Manuel Guijarro, Gavin McCance, Vítor Gouveia
ATLAS    Yuri Smirnov
LHCb      Joel Closier
CMS     Ivan Glushkov

Questions

(JC) Q- Can we have a clear description of what the various images are?
(TB) A- this is in the doc pages for cloud infrastructure
 
(VL) Q- is IT/CS aware that the availability zones should be independent? E.g. for a router upgrade, different availability zones can be affected at the same time if this is not taken into account.
(TB) A- indeed, interventions should be arranged to take this into account but this is not defined with CS yet.
 
(IR) Q- critical power for openstack hypervisors?
(TB) A- not at the moment. Relying on having machines both at CERN and Wigner is probably better than relying on critical power.
 
(JI) Q- could we have step 2 of dnslb alias migration done for us as part of the 1^st ticket?
(IR) A- no, we have no control on the IT/CS configuration. There are about 100 DNS aliases in Quattor.
 
(JI) Q- could we have the host certificate automatically include the DNS alias's name, e.g. as SAN?
(GM) A- It's already very easy to do manually with certmgr::certificate

14:00 → 14:05 OpenStack Images

14:05 → 14:20 Availability Zones

14:20 → 15:00 Load Balancing