 |
 |
It should use the most recently updated conditions set in this case. > On Dec 13, 2014, at 9:11 AM, Graham, Mathew Thomas <[log in to unmask]> wrote: > > I _think_ that if you turn off “use2014Gains” it does the right calculation…so thats what I did. The resulting energies are not completely crazy, though I still have some doubts. > > When the ConditionsManager gets the gains, it says it finds 4 records satisfying the criteria"ecal_gains with” run number 2000-onward (run number is set to 2000) ; what does the the hps-java code do in this case? > > >> On Dec 12, 2014, at 6:03 PM, McCormick, Jeremy I. <[log in to unmask]> wrote: >> >> Hi, Matt. >> >> Nathan put new gain values into the conditions database yesterday, but there is no LCSim Driver that uses this value for a proper energy conversion to CalorimeterHit. Though you probably are generating a CalorimeterHit collection in the job, the energies of those hits in GeV aren't correct. (I haven't verified this because I haven't looked at it lately.) That's because EcalRawConverterDriver is being run but it doesn't use the convention described by Nathan for the energy algorithm. This would need to be written according to the formula given in previous emails to the list in order to make CalorimeterHits that make sense for the clustering. >> >> --Jeremy >> >> -----Original Message----- >> From: Graham, Mathew Thomas >> Sent: Friday, December 12, 2014 12:37 PM >> To: McCormick, Jeremy I. >> Cc: Nathan Baltzell; hps-software >> Subject: Re: ECAL calibrations >> >> We aren't seeing ECal clusters in the monitoring...is this due to the gains not being implemented in the hps-java yet? Or something else (like the clustering drivers not being run)? >> >>> On Dec 12, 2014, at 2:23 PM, McCormick, Jeremy I. <[log in to unmask]> wrote: >>> >>> Hi, Nathan. >>> >>>> Ok, it seems to work now (but spits out lots of info about svt, >>>> severe error ...) >>> >>> Good to hear it is working. The SVT errors can be safely ignored and are being trapped by the conditions manager. >>> >>>> Also we need a command line way to print the table currently used for a given run number. E.g. "give me the gains for run 2001". >>> >>> This exists already.... >>> >>> CommandLineTool print -t ecal_gains -r 2000 >>> >>> etc. >>> >>> Check "print -h" for the exact syntax. >>> >>>> Looks like I accidentally "added" the gains 3 times instead of once yesterday, due to all the "errors" reported. Maybe you want to delete the first two. >>> >>> That's not a big deal. Can you try deleting these yourself from the >>> mysql client? I would like to double check that hpscalibrations has >>> permissions to do this. (It is supposed to have the delete grant.) >>> >>> It would just be a simple SQL command from the terminal like... >>> >>> mysql> delete from ecal_gains where collection_id = 123; >>> >>> You just need to put in the collection ID there that you want to delete. >>> >>> Today I'm going to look at extracting run number from the EVIO events in the new data you put out there so the monitoring app can initialize conditions properly without needing a user run number.... >>> >>> --Jeremy >>> >>> -----Original Message----- >>> From: Nathan Baltzell [mailto:[log in to unmask]] >>> Sent: Friday, December 12, 2014 10:47 AM >>> To: McCormick, Jeremy I. >>> Subject: Re: ECAL calibrations >>> >>> I checked, and the peds and gains in the db look as I expect. >>> -nathan >>> >>> >>>> On Dec 12, 2014, at 9:15 AM, Nathan Baltzell <[log in to unmask]> wrote: >>>> >>>> Ok, it seems to work now (but spits out lots of info about svt, >>>> severe error ...) >>>> >>>> You should find now gains (collection_id=3) and pedestals >>>> (collection_id=4) in the db, but I'm about to run a full test, >>>> downloading them crosscheck that they are what I expect before announcing on the mailing list. >>>> >>>> Looks like I accidentally "added" the gains 3 times instead of once >>>> yesterday, due to all the "errors" reported. Maybe you want to delete the first two. >>>> >>>> Also we need a command line way to print the table currently used for >>>> a given run number. E.g. "give me the gains for run 2001". >>>> >>>> -nathan >>>> >>>> >>>> >>>> >>>>> On Dec 11, 2014, at 5:52 PM, "McCormick, Jeremy I." <[log in to unmask]> wrote: >>>>> >>>>> Hi, >>>>> >>>>> Okay, there was yet another bug. (Conditions system wasn't >>>>> initialized properly.) >>>>> >>>>> But this should work now on trunk... >>>>> >>>>> java -cp hps-distribution-3.1-SNAPSHOT-bin.jar >>>>> org.hps.conditions.cli.CommandLineTool -p connection.properties -v >>>>> load -t ecal_calibrations -f ecal_calibrations.txt >>>>> >>>>> Just pull it again from Nexus once the build is done. >>>>> >>>>> Sorry, I need to test this better after I change things.... >>>>> >>>>> -Jeremy >>>>> >>>>>> On Dec 11, 2014, at 2:29 PM, Nathan Baltzell <[log in to unmask]> wrote: >>>>>> >>>>>> java -cp hps-distribution-3.1-20141211.202153-123-bin.jar >>>>>> org.hps.conditions.cli.CommandLineTool -p >>>>>> hpscalibrations_jlab.properties load -t ecal_calibrations -f >>>>>> ped4db.txt >>>>>> >>>>>> http://www.jlab.org/~baltzell/HPS/ped4db.txt >>>>>> >>>>>> (null pointer at line 483 in DatabaseConditionsManager.java) >>>>>> >>>>>> -nathan >>>>>> >>>>>> >>>>>>> On Dec 11, 2014, at 5:19 PM, "McCormick, Jeremy I." <[log in to unmask]> wrote: >>>>>>> >>>>>>> Oh and I meant to say please attach the ped.txt file in your reply as well... >>>>>>> >>>>>>>> On Dec 11, 2014, at 2:18 PM, McCormick, Jeremy I. <[log in to unmask]> wrote: >>>>>>>> >>>>>>>> Can you send again the exact command you're using and attach the file you're trying to load? I'll test it on my local dev db... >>>>>>>> >>>>>>>>> On Dec 11, 2014, at 2:08 PM, Nathan Baltzell <[log in to unmask]> wrote: >>>>>>>>> >>>>>>>>> Jeremy, >>>>>>>>> >>>>>>>>> Ok, the "add" of gains worked now. >>>>>>>>> >>>>>>>>> But now using latest snapshot (20141211.202153-123) to "load" >>>>>>>>> the pedestals I get an error similar to the one yesterday: >>>>>>>>> >>>>>>>>> DatabaseConditionsManager.findTableMetaData(DatabaseCondisiontMa >>>>>>>>> n >>>>>>>>> ager.java:483) >>>>>>>>> >>>>>>>>> Something wrong with these options:? >>>>>>>>> load -t ecal_calibrations -f ped4db.txt >>>>>>>>> >>>>>>>>> -Nathan >>>>>>>>> >>>>>>>>> >>>>>>>>>> On Dec 11, 2014, at 3:10 PM, "McCormick, Jeremy I." <[log in to unmask]> wrote: >>>>>>>>>> >>>>>>>>>> Hi again, Nathan. >>>>>>>>>> >>>>>>>>>> Okay, this should work fine now. I tested on my local dev database to make sure. >>>>>>>>>> >>>>>>>>>> Executed this command similar to yours... >>>>>>>>>> >>>>>>>>>> java -cp ~/.m2/repository/org/hps/hps-distribution/3.1-SNAPSHOT/hps-distribution-3.1-SNAPSHOT-bin.jar org.hps.conditions.cli.CommandLineTool -v -p ../hps-java-sandbox/conditions_dev.prop add -t ecal_gains -k ecal_gains -c 3 -r 2000 -e 9999 -u baltzell -m 'CosmicGains' >>>>>>>>>> >>>>>>>>>> You should see something like this in the log which indicates it succeeded... >>>>>>>>>> >>>>>>>>>> Thu Dec 11 12:05:01 PST 2014 :: DatabaseConditionsManager :: >>>>>>>>>> FINE :: INSERT INTO conditions ( run_start, run_end, >>>>>>>>>> table_name, name, collection_id, created_by, notes, created ) >>>>>>>>>> VALUES ( '2000', '9999', 'ecal_gains', 'ecal_gains', '3', >>>>>>>>>> 'baltzell', 'CosmicGains', STR_TO_DATE( '2014-12-11 12:05:01', >>>>>>>>>> '%Y-%m-%d %H:%i:%S' )) >>>>>>>>>> >>>>>>>>>> You will want to 'svn up' and build trunk and then use hps-distribution from there. >>>>>>>>>> >>>>>>>>>> Or you can wait for this build to finish... >>>>>>>>>> >>>>>>>>>> http://srs.slac.stanford.edu/hudson/view/HPS/job/hps-modules/83 >>>>>>>>>> 5 >>>>>>>>>> / >>>>>>>>>> >>>>>>>>>> And then download the 3.1-SNAPSHOT bin jar from Nexus. >>>>>>>>>> >>>>>>>>>> -Jeremy >>>>>>>>>> >>>>>>>>>>> On Dec 11, 2014, at 11:15 AM, McCormick, Jeremy I. <[log in to unmask]> wrote: >>>>>>>>>>> >>>>>>>>>>> That shouldn't be a fatal error, as it is being trapped in the >>>>>>>>>>> manager. It just means SVT conditions aren't in the db yet... >>>>>>>>>>> >>>>>>>>>>>> On Dec 11, 2014, at 11:08 AM, Nathan Baltzell <[log in to unmask]> wrote: >>>>>>>>>>>> >>>>>>>>>>>> I just now used the latest jar from nexus, and now it complains about "svt_channels" >>>>>>>>>>>> when trying to "add" a table for "ecal_gains". This is using >>>>>>>>>>>> the same command as yesterday. >>>>>>>>>>>> -Nathan >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>>> On Dec 11, 2014, at 1:28 PM, "McCormick, Jeremy I." <[log in to unmask]> wrote: >>>>>>>>>>>>> >>>>>>>>>>>>> Hi, Nathan. >>>>>>>>>>>>> >>>>>>>>>>>>> I know we discussed before that the currently used calibrations (pedestal + noise) in the JLAB database are wrong for about half the crystals due to some kind of mix up with the channel ID convention that was used. I think we should fix this ASAP so that the calibrations are reasonable. The existing conditions collection should just be deleted, too, as it isn't usable. >>>>>>>>>>>>> >>>>>>>>>>>>> I have an LCSim Driver that generates a pedestal + noise >>>>>>>>>>>>> conditions set from data. It is relatively simple right now >>>>>>>>>>>>> and just uses the mean and sigma distribution of all ADC >>>>>>>>>>>>> values from a histogram. But it could be made more >>>>>>>>>>>>> sophisticated by, for instance, selecting only ADC values >>>>>>>>>>>>> from background "hits". Could we start working from this >>>>>>>>>>>>> Driver to generate these calibrations? If we need to >>>>>>>>>>>>> improve it then we should do it. It is already in SVN. >>>>>>>>>>>>> (mentioned in my talk today) >>>>>>>>>>>>> >>>>>>>>>>>>> Also, are the pedestals sufficiently stable that we could use one calibration for all the runs thus far? I have calibrated just a few runs with this Driver, and I don't have much idea if there is any pedestal drift (none that I have seen), or at least enough for us to want to store this information by individual run. I know you stated that there is no significant amount of pedestal drift. Has this been shown definitively in any analysis/plots e.g. for the cosmic runs? >>>>>>>>>>>>> >>>>>>>>>>>>> I currently have all data files for runs 2823, 2825 and 2826 copied to SLAC where I can do analysis on them. Do you want me to generate the pedestal calibrations for these runs and send them to you as text files so you can verify and then load yourself OR I just can directly load them into the JLAB database myself if you're too busy with ECAL operations. Probably if I did this I would just assign the collections to the individual run numbers and not a range. >>>>>>>>>>>>> >>>>>>>>>>>>> Anyways, let me know what you think is best... >>>>>>>>>>>>> >>>>>>>>>>>>> -Jeremy >>>>>>>>>>> >>>>>>>>>>> ############################################################## >>>>>>>>>>> # >>>>>>>>>>> ######### >>>>>>>>>>> Use REPLY-ALL to reply to list >>>>>>>>>>> >>>>>>>>>>> To unsubscribe from the HPS-SOFTWARE list, click the following link: >>>>>>>>>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTW >>>>>>>>>>> A >>>>>>>>>>> RE&A=1 >>>>>>>> >>>>>>>> ################################################################# >>>>>>>> # >>>>>>>> ###### >>>>>>>> Use REPLY-ALL to reply to list >>>>>>>> >>>>>>>> To unsubscribe from the HPS-SOFTWARE list, click the following link: >>>>>>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE >>>>>>>> & >>>>>>>> A=1 >>> >>> ###################################################################### >>> ## >>> Use REPLY-ALL to reply to list >>> >>> To unsubscribe from the HPS-SOFTWARE list, click the following link: >>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1 > ######################################################################## Use REPLY-ALL to reply to list To unsubscribe from the HPS-SOFTWARE list, click the following link: https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1