Hi Ian, Steve, et al, On Thu, 11 Apr 2002 22:33:39 +0200 "Ian Bearden" <bearden@nbi.dk> wrote concerning "RE: calibration and reconstruction guide": > Hi Steve, et al., > As I understand it, the idea is not to have [any?] local DBs, but one > central. Indeed. However, we have the _option_ for local databases for testing use _only_! This feature, by the way, was what saved us when pii3 had to be taken down, and we could mirror the run database to rcas0005. It's also what has saved me many a nights, because I could mirror every BRAHMS database to a local database, and use that when pii3 or rcas0005 `disappears' for one reason or another. It wouldn't be a bad thing to setup slave servers, so that we had some redundancy. Also note, that there are interfaces for storing the databases in ROOT files, for example using the rootd remove file server daemon. The idea is that you can extract parts of the database to some ROOT files, commit calibrations to those ROOT files, test the calibrations, using the ROOT files, and when you're happy, commit the stuff to the canonical database. All this this ofcourse explained in `The Guide' [2], and the class documentation. > There is no risk of a bad calibration, because the people who do the > calibration will always make sure that the calibration is good > *before* committing it to the DB. Also, the design of the DB is such that calibrations are never deleted - new revisions of the calibrations are added, and take propority over the old ones, unless you mark it explicitly not to. So even if you commit a `bad' calibration, you can still retrieve the old `good' calibration, down prioritiese the `bad' calibration and so on. Note, that this doesn't mean that one shouldn't check the calibrations first, or flood the database with revisions. The point is, that there's a limited amount of storage avaliabe for the revision (2GB for _all_ revisions), so one needs to be a little careful there. In danger of sounding repetitious, All this is explained in painful details at [1,2]. > So far this has not been much of a problem since only Djamel has > done calibrations. That's unfair. Claus, David and Marco also did a lot too. > Now that the instructions are available, others should also pitch in > (and check their results, of course.). > Ian Steve, how you proceeding on making `Br[Mult|Tile|si]Calibration' compatible with `Br[Mult|Tile|si]TmpCalibration' so that we can remove thos old annoying ASCII files and put it all in the database? Radeck and Pawel, all those ASCII files for the DCs, couldn't they be commited to the database? I think we need an executive descision that state descively that after a certain date (say July 2002) _all_ retrivial of calibrations _must_ be via the database, and ASCII files should _only_ be used for testing. Comments anyone? > > -----Original Message----- > > From: owner-brahms-l@bnl.gov [mailto:owner-brahms-l@bnl.gov]On Behalf Of > > Stephen J. Sanders > > Sent: 11. april 2002 22:18 > > To: brahms-l@bnl.gov > > Subject: Re: calibration and reconstruction guide > > > > > > Hi Djamel, > > Thanks for the detailed instructions. This type of "how-to" is > > extremely useful > > to those of us without local "experts" to rely on. > > > > One question: When you talk about commiting to the database, are > > you working with the main brahms DB, or a local copy? If you > > changing the main DB, isn't there a risk of someone messing up a good > > calibration that is already there? > > > > So far I have only "taken" from the DB so I'm not sure of > > protocols for modifying it. Hope that clearifies things a bit. Yours, Christian Holm Christensen ------------------------------------------- Address: Sankt Hansgade 23, 1. th. Phone: (+45) 35 35 96 91 DK-2200 Copenhagen N Cell: (+45) 28 82 16 23 Denmark Office: (+45) 353 25 305 Email: cholm@nbi.dk Web: www.nbi.dk/~cholm [1] http://www.sdcc.bnl.gov/brahms/private/database/index.html [2] http://pii3.brahms.bnl.gov/~brahmlib/brat/guide/bratguide.html#SECTION00090000000000000000
This archive was generated by hypermail 2b30 : Fri Apr 12 2002 - 08:03:50 EDT