Update managing-files-with-ganga.md #258
Conversation
| ```python | ||
| j.backend.getOutputDataAccessURLs() | ||
| ``` | ||
| which will return a list of the PFNs for any DiracFile object created in your job output. |
There was a problem hiding this comment.
(I don't think this is a big issue but) this duplicates the other lesson from first-analysis-steps https://lhcb.github.io/starterkit-lessons/first-analysis-steps/ganga-data.html
Maybe some cross-references could be useful.
|
If I remember rightly (and I probably don't because it was a looong time ago), we figured that CERN replicas were easiest for the analyst because the access pattern was simple. Generating a Grid certificate was annoying in certain environments (e.g. a Docker container, or a CI). But I remember some work on this area a while ago, to try to make things easier. @cburr you looked in to simpler certificate generation right? |
If I could get a valid grid certificate in my CI job environment which grants read-only access to my files, I'd gladly use that over replicating to EOS and using kerberos to authenticate. Currently though I'm not so sure whether it's a good idea to paste an unencrypted grid certificate into a protected/masked GitLab CI variable. |
You have the wrong Chris Burr. You are looking for @chrisburr |
Oops, very sorry! 🙊 (cburr is his username in our internal system.) |
5 participants
I am not sure replicating all your outputdata to CERN is right. You should access files directly with the PFNs.