Computer Center News: (Richard) The cooling outage was close to a non-event for us. SCCS did a good job of isolating us. We were without u05 (ftp area). There were some problems with norics, perhaps related to people accessing unavailable nfs disks in their login scripts. There was an unintended outage of afs01 for an hour or so.
We're asking SCCS to trigger the 2009 order of 250 Tbyte of disk.
Overlay Code is ready to go. Procedure to make requested large quantity of overlay events is in development. (Leon) [summarizing his remarks at the meeting:]
- Run Gleam jobs to skim Periodic-Trigger [PT] events and write out the Overlay file directly, skipping the intermediate, and very large, digi file. We should also write out the merit files, since this will allow us to do some studies on the PT events.
- Examine the merit files from the PT events to determine a reasonable set of McIlwain bins.
- Read the generated overlay files into a root macro, and separate into output files corresponding to ranges of McIlwain-L.
Input needed: It would be useful if someone could identify one or two 55-day cycles of stable running; i.e., without major glitches, pointed observations, etc. We could start with one cycle and add another later.
[Status as of 1/16 from Leon:]
- The Gleam job exists and has been somewhat tested. Unfortunately, each run takes 1 hour+ to process. Most of this time is spent reading in digi events which are then not written out. A 55-day orbital cycle contains about 850 runs and ~7M periodic trigger events. Philippe is suggesting twice that number. So we need some combination of job speed-up and pipeline running to get through the ~1700 runs. (Tracy says that rewriting digiRootReaderAlg to use converters would probably fix this problem, as then we could read in only the header and then read the rest of the digi only if the event passes.
- The root macro that examines the PT events and assigns McIlwain ranges is written and needs a bit of testing. The current scheme divides the range up into bins of ~equal numbers of events except for the higher values of McIlwain-L; there the bins are finer because the ghost rate is higher.
- The skeleton of the macro that reads in Overlay events and writes out "binned" events exists and works. Thanks for the help, Heather!
(Philippe) How does OnboardFilter function with overlayed events? (Tracy) Simulated part of the event includes trigger bits as always. Overlay comes with trigger bits from the data. OBF operates on the OR of these.
(Philippe) would like to see a GR release including LPM and interacting alpha along with overlays. (Heather) There is no special difficulty in doing this; she's waiting on blessing for interacting alpha code.
ACD upgrade (Richard) Eric Charles has been working on various changes to ACD code. There will be a mini-review next week once Eric is ready and the usual reviewer suspects have been rounded up.
AAS Report: (Richard) With about 2500 attendees, this was the largest meeting he had ever attended. It took over the entire convention center. There was an exhibition area with booths (including one for Fermi, naturally). The other half of that space was used for the plenary talks, with unfortunate results acoustically. Steve gave a fine plenary talk; attendance was hurt somewhat since it was the morning of the last day of a conference ending at noon. There were 2 Fermi parallel sessions. The first was popular; the second was less well-attended since it was at the same time as a GRB session.
ScienceTools: No report.
FSSC Report: Quiet here as well.
Documentation: (Chuck) sends this summary:
I'm still working my way through the flagged items in my email inbox and combing through the workbook, making routine changes as I go. Most of the flagged items have now been incorporated.
Yesterday, I started linking in David Chamont's documentation on the backend of the skimmers, i.e., the line-mode ROOT skimmer known as TSkim. This project has resulted in a fair amount of discussion between Heather, David, and Tom Glanzman regarding where and how best to link it in, and I think we're nearing agreement. At any rate, I plan to have the latest updates posted to the web later this morning.
I also need to create a new tag for the workbook, so Berrie can update the mirror site. And please note that Richard and Heather are currently reviewing my "ToDo" list and priorities for the coming year, so now would be an ideal time for any comments, concerns, or suggestions that people may have.
Inputs for WIRED (Heather) A new GR will be coming which will allow input relation via the event displays. This required modifications to HepRepCorba, HepRepSvc, and RootIo. Updates to allow input merit files via the event displays is coming, just requires a bit more work.
ROOT 5.20 (Heather) ROOT v5.20 plus updated xrootd is built and tested on Linux. Now built on windows with testing to come. A debug windows version will be provided to Tracy.
skimmer (Heather) The latest version of the skimmer is v7r2p1. It would be good to determine what troubles persist with the new version so we can make specific requests of David for further patches.
FluxSvc verbosity (Tom) provided this summary:
Johann observed a plethora of messages like this in his recent MC pipeline runs:
ExposureAlg INFO tick at 257112815.1 sec, lat, lon, magLat, zenithTheta, SAA = 3.8, 60.8, 9.1, 0, 0 FluxSvc INFO entering SAA at 257117145.2
As his runs span many weeks, there are lots of these message. The first message can be reduced in volume by using a job option:
ExposureAlg.PrintFrequency = 10;
(default is '1' which means print every 'tick'). I don't know how to suppress the second message without new code. And I recommended keeping those messages anyhow.
f2c (Joanne) made minimal changes to genericSources to eliminate its dependency (hence also FluxSvc and GR altogether) on f2c. The result works, but leaves something to be desired stylistically. Jim did a better job, now committed and tagged as v1r12p1. There is no easy way to get rid of ScienceTools dependence on f2c, however. FSST now includes f2c as part of its distribution; we will probably have to do likewise.
SCons RM, Installer (Navid) writes:
rhel3, 4, 5 as well as 64bit rhel4 continue to build as expected. Last week's outage shut down our mac machines. They're back up, but queued builds aren't submitting yet. I have requested the admins at SLAC to look into it and see why that is. In the meantime I've been working on getting installer releases working. The new RM creates user releases at the moment (as opposed to developer releases, which will come later) without the dependency on the external libraries recorded. I'm in the process of having that done and then I'll work on creating a download interface.
obf external Emmanuel has a working installation with native organization. See also a proposal Joanne wrote up to use an installed organization along lines suggested by Navid. It would simplify linking and decrease size of the distributed external.
Fred (Emmanuel) now has a suitable FOX library. However Fred is seg faulting when started.
More external matters (Heather) James Peachey has created a new cfitsio patch, V3060A. This is in response to the JIRA STGEN-83. There has been a long standing request for us to migrate to ape 2.4. Both externals are built and have been used in the ST LATEST builds. Heather tested on windows in response to a request from Jim and James. Now all that's left is to generate a new ST using the updated IExternals.
New tagging convention (Heather) When do we start making new style tags? It is straightforward using either GoGui or the command-line application stag. (Joanne) Executables for RHEL3, RHEL4, Windows and Mac can be found in the u05 ftp area.
previous | minutes index | next |