Wildcards in unpacked.cern.ch

We are starting to roll out supports for * wildcard in unpacked.cern.ch and DUCC.
The code is already updated, so we are already ingesting some images using the * glob.


One example is the following from our friends in cms.


This ingest all the tags that start with rhel7-m, so rhel7-m201911, rhel7-m202001 and rhel7-m202002 at the moment.

It is possible to use the wildcard to match also prefix and not just suffixes.

Of course is possible to use the wildcard also to match all the tags of an image.


Internal details

Internally we use a regex engine, we convert * to .* and match the regex to all the tags of the images.
This means that we could also add support for the + wildcard.


From preliminary tests, it seems that everything is working fine, if somebody as an image to ingest it would be very interesting to try that as well.

The relevant PR is this one for the interested: https://github.com/cvmfs/cvmfs/pull/2490

1 Like


Lukas has added all the atlas athena and analysisbase images. It seems to work well. There is now some manpower to produce at least some of the production releases automatically in the future. I thought we had to add something in the process to put them in CVMFS but perhaps this is enough.