Ryan, Matthew, Jonathan, Bruno, Nate, Kanika, Craig, Gavin, Jeny, Paola, Michael, Satish, Raphael, Pavan.
First analysis tag for simulation validation is ready FA14-09-23. Despite only being two days old this release already requires a number of tweaks before is is ready to be used for the complete simulation production. The missing bits are:
- Beam flux files. The v04 ones we've been using until now are old (2011) and need updating, especially given the updated location of the ND. Raphael et al. are making these now, these should take a day, followed by a couple of days for gsimple production. They're currently suffering from low grid priority. When these are done we'll need to change a path in the release to point to them.
- Genie weights are being propagated wrongly, leads to a memory overhead.
- Genie weights were not included in mixing. This needs a fix.
Given all these issues it was decided to make validation files now for genie events and ND rock singles to test that the machinery works. Then to run full scale production on these early next week. The exact schedule will depend on how the validation proceeds, in theory it is fine to run rock singles now with the v04 flux files as they shouldn't be any different. The genie though needs to wait until the new files are ready. The consensus was to re-tag when this is done and use the new release to make proper validation samples and then hopefully also the full production. The cosmic files, however, do not need to wait for new flux files, and seeing as these are conceptually different from genie files and there is a pressing need for them (especially in the ND for calibration) it was decided to run the ND cosmics now as this is fast and also to run a validation sample for the FD cosmics (as this isn't fast) - full FD cosmic production planning will wait for the output of the validation step.
The simulation validation tag production test results (Matthew)¶
docdb-12057. In summary everything worked that was expected to work. However the memory usage is a little high (but not super dangerously high) and there was an odd feature where ND genie detector fails when run in the same job as the mixing. This issue disappears when these are run individually (as production will be run). Also ND genie RHC doesn't work yet as we have no flux files.
New art release (Jonathan)¶
A new ART release has become available since the first, first analysis tag. We discussed whether we should include this in the tag, in development or both. The consensus was that this version isn't important and in fact includes an xrootd bug, so we'll skip it for now. Hopefully later major updates to art won't fall between the creation of validation and proper releases so will fall naturally into our development cycle.
Keepup updates: PC-hits (Paola & Satish)¶
Paola has figured out the problem with the jobs taking 4x what they should to do with mis definition of variable “ROLE”. and some grid database. She has fixed this and informed the group. She will now proceed with taking over FD pchits keep up and starting up the ND keep up.