CTA setup at CERN


We continue testing CTA at DESY and curious how CTA is set up at CERN. For example, are there dedicated Frontends/db servers/tape servers? Roughly how many drives per host and how does it scale?

Puppet infrastructure? Are there configs we could borrow to avoid reinventing the wheel?
General recommendations?

Goes without saying that there’s no one size fits all but this would give us a rough idea on which steps to take.

Many thanks,


Julien is responsible for all Puppet profiles, frontends and EOS part. I myself (subjective) see no reason why we couldn’t share with you a dump of that Puppet configuration as is. It is heavily CERN dependent, but some basic ideas could help you to move forward quicker.

Same goes for various operations tools / wrappers that we use. I can send you Git repo dump as is - no guarantee.

For the tape servers, we have 1-to-1 connection:

tape drive <-fibre-> tape server <-ethernet-> the rest of EOS CTA

(Remark: RAL is using more than one tape drive per server, so ask them for their experience).

If you need more info, please let me know but perhaps if you could come up with list of concrete questions, we could simply cover them in a meeting.

Best regards,


Hi Vlado,

Yes it would be great if you could share w/ us the Puppet config and the operational tools. This would give us a few pointers/rough idea on what we need to extend and what not.

thanks a lot.

I provided DESY with CERN CTA Puppet profiles as well as CTA operations scripts.

I highlighted the fact that it was provided AS IS, no support what-so-ever.


1 Like