Hi
> _ some modification in dst (allow for run #, I think this is why we lrd
> needs to access lrn file)
run number is in DST, together with a few other per run parameters such as
gotime, endtime. The matter is just to read it out.
the reason why lrd has to access lr file is because it has to reproduce lr
ntuple. From DST only tracked events are recoverable while lr contains
every triggered event. Because lr file has to be accessed, some entries
like run # was just taken from lr straightly without the trouble of
reading from DST.
If we decide to ditch lr and mandate lrd not to access lr/flr file then
lrd has to be modified to cope with that.
> _ and hopefully some seg fault fixes. I am still asking your help to
> collect information about that.
Some of the seg-faults in v3.0 was debugged in April and the changes went
into the head. Many of them happens when epics and scaler data stream is
broken. For instance Coda died, restarted, run started yet Epics server is
still initializing. When a physics event could not find an Epics/Scaler
event that matches its time stamp, it could query for the -1th entry of an
arry thus causes seg-fault. A few line of codes are added to the head to
guard against such exceptions and the head was tested on a few seg-faulted
runs. The debugging was not done for all seg-faults because a lot of coda
files was moved from spud4 disc to other places and was not accessible at
that time.
Chi
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:31 EST