Comings and goings: We congratulate Debbie Nicholson on her move to the Particle Physics & Astrophysics Directorate, but alas for us! Those at SLAC can still visit her at the Kavli Building.
Richard, Leon and Toby are at the Bari meeting this week. Tracy will be leaving on Saturday; will return towards the end of the month.
Big run: As always, follow progress on the Big Run Checklist. (Tom) With v13r9p3 we've successfully made over 140,000 runs for OpsSim2, the most ever. There was a hiccup due to an AFS server outage. And Sunday morning an nfs disk ran out of inodes, causing a 6 hour outage. Subsequently there were problems rolling back failed runs. For the future, we'd like to make the Pipeline smart enough to recognize systematic problems like this and throttle itself.
There were two successful all-gamma runs at IN2P3; GRB grid runs are almost ready to go.
Science Tools report: (Jim) talked from these comprehensive notes.
Data handling: (Dan) There were database problems at a time when several activites were in progress concurrently. We're now retrying them one at a time to find the culprit(s).
Given throughput we've seen so far, we expect the Big Run to complete on about Feb. 21.
[Weekly minutes for Data Handling can be found in the usual place. ed.]
Documentation report: (Chuck) has been reviewing the SAS How to Fix page in Confluence. Richard has asked me to make sure it covers all of the major services provided and includes a "recipe" for restoring them to service in the event of an outage. All ideas, suggestions and comments are welcome.
This month's Newsletter is now out on the virtual newstands.
ACD Review (Heather) The second part of the review took place last Thursday. Work to be done is being prioritized. One item high on the list is to implement and take into account screw holes (both top and side). [Geometry currently includes an accurate description of holes in the top; no holes in the sides. Acd code may not be taking the top holes into account fully yet, however.]
(Tracy) is immersing himself in the AcdRecon code in order to understand it fully. Michael and Leon [also on the review committee] will meet in Bari.
Interleave status: (Tracy) Code now checked in runs and makes a ROOT file, but he suspects it isn't doing quite what was intended; he needs to look at CEL again. One possible problem area has to do with file activity. Interleave will open and close files as needed; should confirm that CEL handles this properly. (David C.) plans to add more documentation on CEL this week.
Skimmer: (David C.) has made the input format for Skimmer parameter files more flexible. Comments (indicated with #) and blank lines may be inserted anywhere. The new format is closer to the expected future textual flavor of the CEL format. For details of the changes, see these release notes. For a more general description of Skimmer features and the format of parameter files, see the Skimmer User Guide. The new release (v4r0) will include this improvement; no new use cases yet, though.
The considered new use case was the deletion of some of the input files from the list of files, although keeping the same complete list of events. When searching a given event which is not here because of the deleted files, the skimmer currently consider it is an error, and it is quite easy to make it a simple warning instead. Unhappily, testing this has demonstrated an unexpected behavior from TChain, which is under investigation. This is why this will not be included in v4r0.
Need an application to transform input file paths for Composite Event Lists to their logical form. Such an application should be provided in the near future.
While doing his own debugging, David an unearthed a useful and heretofore undocumented fact about interactive ROOT: one can turn off optimization with the command
.O 0
(dot capital_O space zero)
[Thanks to David for extra detail and documentation pointers. ed.]
SCons status: (Heather) As per Navid's email you can try out the SCons version of Science Tools with the command
cvs co -r ScienceTools-LATEST-1-2220 ScienceTools-scons
(Navid) That tag is now out of date. RM is now making new-style tags for every build, so there should be a new-style tag corresponding to LATEST1.2224.
Now I've shifted to working on the new RM, getting old RM to trigger a new RM build, and getting new RM to put results up on the web. (Heather) is hoping for a more detailed SCons discussion next week, when more people will be in town.
previous | minutes index | next |