sPHENIX HCal discussion

[John Haggerty <haggerty AT bnl.gov>]

Martin,

Great, thanks. I think the event libs have not propagated to the sphenix 
world at rcf, is that right? If that's right and I'm not mistaken, we'll 
want that.

Also, when you get a chance, we're going to need to select the run type 
at the start of every run... I've already written junk when I meant to 
write cosmics and cosmics when I meant to write LED.

On 3/4/16 1:30 PM, Martin Purschke wrote:
> Dear all,
>
> let me add a bit of new information to this.
>
> As John points out, we are taking data with the sphenixdaq machine, on
> the grounds that this is the one that has the jSEB card that is in use.
>
> The data, however, are right now flowing to a disk on the gateway
> machine (highwaygw.phy.bnl.gov), seen as /data/data/phnxsa/...
>
> We are not planning to take this machine to FermiLab -- we then rather
> direct (and copy what's there) the data to a machine called hcaldaq
> (192.168.100.40). We will preserve the mountpoint's name. That machine
> has 6TB of prime and fast disk space.
>
> The reason that data are not going to their final destination yet is
> that JohnH might do some development on that hcaldaq machine that
> requires some reboots.
>
> The elog that existed as an ad-hoc installation on sphenixdaq has been
> moved to hcalgw as well, and is available on port 7815 (the same as the
> 1008 Elogs, the phone nr there is easy to remember).
>
> I would suggest that we all use the much faster hcaldaq machine for
> heavy-duty processing. Eventually the data from the DAQ will be local
> there, and the machine has a lot more memory, 6 CPU cores -- you will
> have a much better experience. The standard account is phnxsa - as in
> "standalone" - the same as on the sphenixdaq machine, but other accounts
> could be made on demand.
>
>
> Now to the cool new things.
>
> The hcaldaq machine has what is now the "sPHENIX" version of the event
> libraries. There has been a lot of cleanup, and there are a few new
> features that I hope you will like.
>
> To begin with, the "-p" switch of the ddump utility now takes a single
> packet id as before, but also id lists and ranges - you can now say, for
> example
>
> ddump -p 1001 ...
> ddump -p 1001,1002 ...
> ddump -p 1001-1005 ...
> ddump -p 1001,1002,3001-3005 ...
>
> Also, reading from a file is now the default, so while the "-f" switch
> is still there, you can omit it.
>
> The "et" functionality is gone, at least for now - it is simply not
> compatible with 64bit. (Did I mention that all is 64bit?)
>
> Instead we have a new feature that can be used in online monitoring. A
> new event stream connects to a monitoring port in rcdaq, and gets events
> from the DAQ opportunistically (since we are not allowed to hold up the
> data taking, we have to tolerate that we miss events).
>
> In dlist and dlist you would say
>
> dlist -r sphenixdaq
>
> and dlist connects to that monitoring stream and gets the next event.
> You will see that rcdaq often takes data without actually logging them,
> and this is a great way to look at a few events, say, to time in
> something, without actually logging them to disk.
>
> In a pmonitor project you open that stream with the command
>
> rcdaqopen ("sphenixdaq");
>
> and can then pstart() your monitoring.
>
> If you happen to be on the local machine where rcdaq is running, you'd say
>
>     dlist -r localhost
>
> and
>
>    rcdaqopen("localhost")
>
> but "localhost" is the default and in both cases you can just say "dlist
> -r" and rcdaqopen();
>
>
> Also, you might want to take a look at the rcdaq manual -
>    http://www.phenix.bnl.gov/~purschke/rcdaq/rcdaq_doc.pdf
>
>
> Best,
>         Martin
>
>
> On 3/3/16 12:35, John Haggerty wrote:
> > This may not be the final word on this, but since we have moved to
> > taking data with rcdaq, we've moved to some new data directories, too.
> >
> > - Taking data is still done on sphenixdaq, which will be difficult for
> > most people to get to (right now, the only way in is via the gateway
> > machine Martin set up in the highbay, called highbaygw.phy.bnl.gov; in a
> > few weeks, that will be a machine at Ferrmilab, so don't get used to it).
> >
> > - The data directory is an automounted NFS volume on highbaygw called
> >
> > /data/data/phnxsa
> >
> > If the automount falls off, you can cd to it
> >
> > - Chris has new place for data at RCF and I made a directory for the
> > beamtest there:
> >
> > /sphenix/data/data01/t1044-2016a
> >
> > and I have been copying data from sphenixdaq there periodically.  We'll
> > try and automate that, but probably I'm more excited about the data than
> > an automaton anyway.
> >
> > There are run-type directories below that called cosmics, led, and data
> > that I'm not wedded to, but right now, we're writing data to cosmics for
> > the most part.


--
John Haggerty
email: haggerty AT bnl.gov
cell: 631 741 3358