From: W.J. Llope (llope_at_physics.rice.edu)
Date: Tue Jan 21 2003 - 15:57:55 EST
At 12:54 PM -0600 1/21/03, Frank Geurts wrote:
> a very unfortunate bug ... i will talk to Herb about this since it
>can be fixed in offline.
i am very glad to read that this can be fixed in offline -
thank you in advance for checking with Herb.
the block of data we've collected so far is not small
enough to simply ignore in my opinion.
one reason i remain confused about this is the fact that
on last saturday you mentioned that you had looked at an actual
daq file from a test run (from Dave h.) in some detail and all
looked good. this is something that started between then?! if
not, how did your reader work if the headers were fouled?!
i.e. this thread implied to me that this something that didn't
break our daq reader but DID break daq's daq reader - i thought
these were (exactly or at least algorithmically) the same....
thank you in advance for any comments
> But, as of now (somewhere after run 4021037) the problem is fixed
>and no more offline tricks are required.
>
>
>
>
>-------- Original Message --------
>Subject: header inconsistency in TOFP (was Re: Some fixes ...)
>Date: Tue, 21 Jan 2003 13:10:01 -0500
>From: Frank Geurts <geurts_at_rice.edu>
>Reply-To: starqa-hn_at_connery.star.bnl.gov
>To: starqa-hn_at_connery.star.bnl.gov
>Newsgroups: http://www.star.bnl.gov/HyperNews-star/get/starqa.html
>
>i looked into the assert() error and found a header inconsistency in the
>TOFP bank (the banklength only included the header length and not the
>full bank length). This has to be (and is) solved in the daq software
>for TOF.
>
>The assert() in the fast-offline discovered this inconsistency and
>seg-faulted the job. Contrary to what I claimed below, this is unrelated
>to the increase in datawords for TOF in this year's run.
>
>To (fast-offline) analyze the data that has been taken so far, I suggest
>temporarily disabling the assert() check for TOFP banks in
>EventReader.cxx. Actually quite a few banks have already exempted
>themselves from these checks!. All data taken as of this very minute
>will have consistent bank headers again and keeping such consistency
>checks proofed to be very helpfull
>
>-frank
>
>Frank Geurts wrote:
>> aha! That could make sense, although there is no assert() in the
>> TOF_Reader it's number of datawords has changed. Let me look into that
>> and I'll check in the new reader today.
>>
>> -frank
>>
>> Jerome LAURET wrote:
>>
>>> I made some fixes to the code today but I also noticed before a new
>>>dev release
>>>that there are remaining problems. Check
>>>
>>>http://www.star.bnl.gov/devcgi/JobInfo.cgi?PT=dev&Trigger=unknown&Status=0&button_name=Submit+Query
>>>
>>>
>>> for example for a list of runs which crashed (for sure) in
>>>fastOffline. This
>>>is caused by an assert() which failed on end<numberOfDataWords ... and
>>>not a
>>>bug "per say" but it sound like we have a potential problem with the
>>>data stream.
>>>
>>> The log says ...
>>>
>>> >> end=7, numberOfDataWords=0, bankname=TOFP.
>>>
>>> for st_physics_4019010_raw_0030015.daq (one example).
>>>
>>> I am leaving now ... but is this persists, this should be checked
>>>carefully ...
>>>In any case, Fastoffline will continue now with the fixes. We shall see.
>>>
>>> Thanks,
>>>
>>
>>
>>
>> -------------------------------------------------------------
>> Visit this STAR HyperNews message (to reply or unsubscribe) at:
>> http://www.star.bnl.gov/HyperNews-star/get/starqa/1054/2.html
>
>
>-------------------------------------------------------------
>Visit this STAR HyperNews message (to reply or unsubscribe) at:
>http://www.star.bnl.gov/HyperNews-star/get/starqa/1054/2/1.html
--W.J. Llope, Ph.D. Res. Assoc. Professor T.W. Bonner Nuclear Laboratory phone: 713-348-4741 Rice University, MS-315 fax: 713-348-5215 6100 S. Main St. Houston, TX 77005-1892 e-mail: llope_at_physics.rice.edu WWW home: http://wjllope.rice.edu/default.html
--
This archive was generated by hypermail 2.1.4 : Thu Jul 24 2003 - 00:39:32 EDT