Metadata from ccin2p3 and gridka was checked (by Mike and Ian) and looks OK except for details of sizeK parameter. This should be corrected in runjob v06.01.00 to round upward. It probably doesn't make much difference of this is corrected in any existing metadata. Rounding off < 1K would probably only have an impact if a large number of such files were in a nearly full cache.
Proposal was made to use same naming convention for merged files from remote sites as is used on fanl farm. I.e. name would have the form:
recoST_reco_all_0000176852_mrg_firstpartition-lastpartition.raw_p13.06.01_000_p14.05.01_ccin2p3
This matches the naming pattern already in use and guarantees a unique file name. The difference will be that the first-last partition will not imply that all partitions in between are present. This might cause some confusion if people are not aware of it since one could find files like ...mrg_004-012... and also ...mrg_005-017... Both would be fine and do not overlap, but that would not be obvious from the name.
It was requested that a metadata summary and naming conventions be accessible on the web. We will put this stuff somewhere in the reprocessing page. Kors had also sent around mail with a request that other procedures be posted as well. We will try to do that as soon as time permits. [Comment: The naming conventions are written in the previous minutes.]
Philip Lewis is working on getting recocert going on thumbnails transferred to FNAL. Some problems have shown up in the current release. Philip s in contact with Harry and Suyong to work things out.
Philip will use the p14.05.00 thumbnails that have been transfered to get things going, but we need to rerun the certification with p14.05.01 since the selection has changed in that version.
That also means we need new master event lists from Marco. These sould be posted on the web also.
Supposedly selection is the only change in p14.05.01 so Philip should still be abel to use the unselected p14.05.00 thumbails produced on the fnal farm as standards.
Need to give Philip aaccess to WWW account so he can put plots on web.
Situation with storage of DSTs is still a bit uncertain. Some site can store the DSts temproraily until they can be stored at fnal, but not everyone has storage capacity for them.
Need to get the auto-destinations maps updated for all the remote sites to allow them to properly store the DSTs. Mike Diesburg will talk to Carmenita Moore to maek sure the maps are setup.
The lasrt CPB meeting left the DSRTT storeage issue in a slightly ambiguous state. It was agreed by all that it would be good to save the DSTs, but tape resources had to be identified for the storage. Greg agreed to identify old data sets tha could be deleted to free up space.
Late Note from this morning: I talked to Amber about this. The effort is ongoing to find tapes, but it will take a while to finish. We agreed that the sites which cannot store the DSTs locally should go ahead and store them at fnal under the presumption that resources will be found.
There has been a re-examination of the processing priority in light of the good performance of p14.05. The new run priority list will require new processing projects for the remote sites. These will be posted off the reprocessing web page. The page gets updated daily. New projects should be visible by tomorrow.
Sites should not use the old projects. Note that he old projects cannot be deleted from sam if they have already been accessed, so they will still show up in listings.
GridKa will redo certification dataset with p14.05.01 and then start with reprocessing. Lyon can start with reprocessing. Both sites should use runjob v06.01.00 and make sure that the application version is p14.05.01 for production.
Phil will test the TMBs from these sites in parallel.