Project

General

Profile

Running DAQ Interface » History » Version 183

« Previous - Version 183/223 (diff) - Next » - Current version
John Freeman, 05/05/2015 12:38 PM


Getting help

If, after reading this document (in particular, the "Troubleshooting and FAQ" section) you run into issues, or have questions, please contact:

Logbook

Please document all changes, software updates, and operational activities to the ELog:

http://dbweb3.fnal.gov:8080/ECL/lbne_35t

Brief preliminaries

This guide assumes you have access to the lbnedaq account on the LBNE 35 ton gateway node, lbne35t-gateway01.fnal.gov . It also assumes you have a rudimentary knowledge of how to navigate a bash environment in Linux (how to change directories, log onto different machines, perform time-ordered listings of files, edit a file using emacs or vi, etc.).

The installed code is located at lbne35t-gateway01.fnal.gov:/data/lbnedaq/daqarea/lbnerc. If you need access to the lbnedaq lbne35t-gateway01.fnal.gov account, please contact one of the people listed above under "Getting Help". Once you have access, to get started, login and first setup the environment:

ssh lbnedaq@lbne35t-gateway01.fnal.gov
cd /data/lbnedaq/daqarea
source fireup

This will take you into the /data/lbnedaq/daqarea/lbnerc directory, out of which the DAQ is controlled. For the remainder of this wiki, the assumption is that you've already setup the environment using the above commands.

Finally, when you see an ellipse ("…") in example output, that is to be interpreted as a placeholder for output which is considered irrelevant to the point being explained.

Killing, launching, and getting the status of the DAQ applications

The DAQ application suite consists of RunControl, the configuration manager, DAQInterface (the component which intermediates between RunControl, the configuration manager, and the artdaq processes) and the XML-RPC server for the TDU (the program which allows a sync pulse to be sent to the hardware at the start of DAQ running). The following points about these programs should be observed:
  • These applications all need to be running in order for the DAQ to work
  • DAQInterface needs to be launched after RunControl, and unlike the other applications, if a DAQInterface process is already found to be running on login, it shouldn't be used, but rather, killed and restarted (unless someone is actively using the DAQ already, of course; a good way to check this is to see when the last time a logfile in /data/lbnedaq/daqlogs/pmt was modified via "ls -ltr /data/lbnedaq/daqlogs/pmt").
  • The actual names of the processes are "lbnecontrol" (RunControl), "daqinterface" (DAQInterface), "CfgMgrApp" (the configuration manager) and "tdu" (the TDU XML-RPC server)

Getting the status

In order to determine which DAQ applications are already active, two approaches can be taken, depending on whether or not RunControl is itself active. If it isn't, you'll need to enter the following:

check_daq_applications.sh

which will produce output which looks like this:

lbnecontrol: Available
CfgMgr: **Not Found**
DAQInterface: Available
TDUControl: **Not Found**

Most recent DAQ logfile written (current US Central Time is Apr 10 09:56):
Apr 8 17:25 /data/lbnedaq/daqlogs/pmt/pmt-32342.1-20150408171225.log

In other words, the above command will both tell you which, if any, of the four primary DAQ applications are running, and then tell you the time of the most recently modified logfile; this last piece of information is helpful in determining whether the DAQ is currently in active use by another individual. In particular, you see that RunControl and DAQInterface already exist, but the configuration manager and the TDU's XML-RPC server do not. In this case, you would not only want to launch the configuration manager and the server, but if the existing instance of DAQInterface hadn't been been launched by you in your terminal, you would also want to kill and relaunch DAQInterface-- an action which should be safe in the context of this example, since you can see above that it's been almost two days since any DAQ output was logged.

Another way of checking which applications are alive is through "lbnecmd check"; note this command will only work if RunControl (lbnecontrol) is itself alive:

lbnecmd check

You'll then see output similar to that of check_daq_applications.sh as far as which applications are alive; however, it contains additional information not shown which will be discussed later:

lbnecontrol: Available
CfgMgr: **Not Found**
DAQInterface: Available
TDUControl: **Not Found**
...

Be aware that if you attempt "lbnecmd check" when RunControl is not yet available, you'll simply see "check failed: '[Errno 111] Connection refused'. Is lbnecontrol running?".

Cleanup: killing RunControl, DAQInterface, the configuration manager, and the TDU XMLRPC server

As of March 30, 2015, a new script called "kill_daq_applications.sh" can be run at the command line to kill the DAQ applications. In order to kill all four applications, one would run:

kill_daq_applications.sh -c -d -r -t

where each option corresponds to the killing of a particular application: "-c" for the configuration manager, "-d" for DAQInterface, "-r" for RunControl, and "-t" for the TDU XML-RPC server. If only a subset of these four applications are running, then only use the relevant options (e.g., simply run "kill_daq_applications.sh -t" if the only application running is the TDU server). Running the script without any options at all will print out instructions similar to what you see here. Be aware that it's good etiquette to make sure someone's not currently using the DAQ before you start killing applications.

For the previous instructions for killing the applications (still correct, though more work-intensive), click on "Show", below:

Launching RunControl, DAQInterface, the configuration manager, and the TDU XMLRPC server

Also as of March 30, 2015, a new script called "launch_daq_applications.sh" can be run at the command line to launch the DAQ applications. In order to launch all four applications, one would run:

launch_daq_applications.sh -c -t -r -d <daqinterface configuration file>

where unless you have your own working copy of the DAQInterface configuration file (what this is will be described later in this wiki), "<daqinterface configuration file>" should be "docs/config.txt". The options refer to the same applications as in the case of kill_daq_applications.sh, and like kill_daq_applications.sh, executing the script without any options will print out instructions. Be aware, however, that DAQInterface will not be launched if RunControl isn't already running. In practice, though, you can launch RunControl in the same launch_daq_applications.sh command by preceding the DAQInterface option with the RunControl option -- so, if the configuration manager and the TDU server are already up and available, then one would only need to execute "launch_daq_applications.sh -r -d <daqinterface configuration file>" in order to have the full suite of DAQ applications available for use.

Note that if an application of a given type is already running, launch_daq_applications.sh will not launch a duplicate application, but will instead print a warning to the screen.

For the previous instructions on how to launch the DAQ applications (still correct, but more detailed and labor-demanding), click on "Show" below:

Performing a Run

With the applications launched, you can take the DAQ system through the standard transitions to perform a DAQ run. The "ground state" of the DAQ is called the "stopped" state. Each successful transition brings the DAQ into a different state. Described in more detail later, the chain of states (in quotes) and transitions (in italics) in the DAQ system can be represented as follows:


"stopped" -> init -> "ready" -> start -> "running" -> pause -> "ready" -> resume -> "running" -> stop -> "ready" -> terminate -> "stopped"


Checking the state of the DAQ

At all stages, you can use the "lbnecmd check" command to see what state DAQInterface is in; example output of this is as follows:

lbnecontrol: Available
CfgMgr: Available
DAQInterface: Available
TDUControl: Available

Run number: 797
Run configuration: ganglia_test
Run type: Test
daqint@localhost:5570 (synchronous): running

Here, you've already seen how lbnecmd check can tell you whether a given application type has been launched or not; on the bottom line, however, you'll also see a description of the state of the DAQ -- in this case, "running", meaning that it's actively acquiring data.

Note that you'll want to wait for a transition to complete before issuing another one, otherwise you'll receive a warning and the transition request will be ignored. DAQInterface will also report when transitions are complete -- e.g., at the end of the "initialize" transition, you'd see:

Initialize transition complete; if running DAQInterface in the background, can press <enter> to return to shell prompt

If a high "debug level" has been set for DAQInterface (see below for more on this), this message can get buried in a set of output messages; in this case "lbnecmd check" may be the easiest way to determine what state the system is in.

Selecting a run configuration and DAQ components

Before initializing or starting a DAQ run, a configuration and set of DAQ components must be selected. This is done via RunControl.

List and select a configuration:

To see the list of available configurations (queried from the configuration manager):

lbnecmd listconfigs

This lists all available configs, as well as the current selected config (Note, default config is "No Config"):
Available configs (Name : description)

8xssp_with_nova_sync : Sorry, no description found.
6xssp_ext : Sorry, no description found.
2xssp_ssp01_led_ssp02_tstamp : Sorry, no description found.
demo : This is a demo.  Testing 1, 2, 3…
...

Current selected config: No Config

You can then select a configuration -- for the purposes of this tutorial, let's go with "demo", which creates simulated data without using any hardware:
 
lbnecmd setconfig demo

This will return "OK" if successful; "lbnecmd listconfigs" will then show this as the current config. If the specified configuration is not known, an error is returned. Instructions on how to add a configuration are given later in this document.

List and select DAQ components.

You can select which portions of the LBNE DAQ setup you want to use in the upcoming run. These are mapped generally to the ArtDAQ BoardReaderMain processes
that read them out. As an operator, you need to ensure that there are configuration files available in the configuration manager for the DAQ Components that you select here, or DAQInterface will return an error.

To see the list of available DAQ components:

lbnecmd listdaqcomps

Will return a list of components available and selected (Default is ALL available components), e.g.:
Available:
component01 (lbnedaq2:5205)
component02 (lbnedaq2:5206)
…
ssp08 (lbnedaq1:5214)

Selected:
component01 (lbnedaq2:5205)
component02 (lbnedaq2:5206)
…
ssp08 (lbnedaq1:5214)

The components are shown by name, along with the requested host/port to run the component's corresponding BoardReaderMain process on. You can then select a list of DAQ components to use in the run:
lbnecmd setdaqcomps component0{1,2}

(n.b. the {1,2} in brackets is a Linux bash shell trick which expands component0{1,2} to component01 component02). This will return "OK" if successful ("lbnecmd listdaqcomps" will show this in the "selected" set of components). If the specified components are not known, an error is returned.

How to make a new component available to RunControl is described later in this document.

Initializing, starting, stopping and terminating a Run

For each of these transitions, please recall that while "lbnecmd <cmd>" returns very quickly, the actual issued command can take several seconds or minutes to complete. Before issuing another command, be sure that the transition is complete by using the check command:

lbnecmd check

For example, before issuing the "start" transition, you can use this command to ensure that the DAQ is in the "ready" state and not still in the "initializing" stage.

First, take DAQInterface from the "stopped" to the "ready" state; this will create the artdaq processes on the hosts they've been assigned, and initialize them with the FHiCL documents:

lbnecmd init daq

Take DAQInterface from the "ready" to the "running" state, in order to begin taking data:

lbnecmd start daq

Immediately, you'll see a request for a confirmation that the configuration and selected components are in fact the ones you want; it will look something like the following:
DAQ config
**********
Run Type: Test
Selected config: ganglia_test
Selected DAQ components:
   ssp01 (lbnedaq1:5207)
   ssp02 (lbnedaq1:5208)
   ssp03 (lbnedaq1:5209)
Start DAQ with these settings? [y]|n: y

Assuming you're happy with the configuration and components, hit "y" to proceed, and as with the other transitions, wait for the "transition complete" message to appear.

Note that the current run number is displayed when DAQ is in the running state using the "lbnecmd check" command. If the start transition was successful, information about the run is saved in the run records directory, /data/lbnedaq/run_records/<run_number>, where <run_number> is to be taken as a stand-in for the actual run number. More on how to understand the directory's contents is described below, under "Examining the output". Of immediate interest, the real-time output of the DAQ will be saved in a logfile located in /data/lbnedaq/daqlogs/pmt/. While in the running state, you can obtain its name just by executing the "check_daq_applications.sh" script described above; with its name, if you open up a fresh terminal and run "tail -f <logfilename>" you'll see the logfile updating in real time.

Pause the running of the DAQ, putting DAQInterface into the "paused" state. This will cause the current open output Root file, in /data/lbnedaq/data, to be closed and given a name reflecting the current run and subrun numbers (the format as of April 10, 2015 is lbne_r<run number>_sr<subrun number>_<creation time>.root):

lbnecmd pause daq

Resume the running of the DAQ, returning it to the "running" state. This will restart data taking to a new output file:

lbnecmd resume daq

Halt the running of the DAQ, returning DAQInterface to the "ready" state from the "running" state:

lbnecmd stop daq

From the ready state, you can start a new run with the same configuration and DAQ components you originally selected.
NOTE To select a new configuration or DAQ component set, you must issue the "terminate" command, which will kill all ArtDAQ processes and return DAQInterface to its "stopped" state:

lbnecmd terminate daq

Examining the output

Once you've completed your run, there are typically two primary outputs: the Root files which contain the data which passed through the DAQ (saved, as of March 25, 2015, in /data/lbnedaq/data on the gateway node), and a set of text files containing information about the run. To access these text files, you'll want to go to the run records directory, which defaults to /data/lbnedaq/run_records/<run_number>. This directory contains the FHiCL documents used to control the ArtDAQ processes in the run, the DAQInterface configuration file used (described later in this wiki), and a metadata file, which, among other things, describes where to find a record of the DAQ's output. An example of this file, for run 778, can be found if we look in /data/lbnedaq/run_records/778 at "metadata_r778.txt":

Config name: demo
Component #0: component01
Component #1: component02
lbne-artdaq commit: 0f0f9be1c63a3b487170579c887ce79944eca6f8
lbnerc commit: 92935f2702c5e0aa732fd3cc9ff6758e1a0c288c
/data/lbnedaq/config/ commit: 97d8b222742a41d5d10ea942a063ddfacac3ce93

pmt logfile(s): lbnedaq3:/data/lbnedaq/daqlogs/pmt/pmt-25771*.log

boardreader logfiles: 
lbnedaq3:/data/lbnedaq/daqlogs/boardreader/boardreader-20150320163004-lbnedaq2-37365.log
lbnedaq3:/data/lbnedaq/daqlogs/boardreader/boardreader-20150320163004-lbnedaq2-37366.log

eventbuilder logfiles: 
lbnedaq3:/data/lbnedaq/daqlogs/eventbuilder/eventbuilder-20150320163004-lbnedaq2-37364.log
lbnedaq3:/data/lbnedaq/daqlogs/eventbuilder/eventbuilder-20150320163004-lbnedaq2-37363.log
lbnedaq3:/data/lbnedaq/daqlogs/eventbuilder/eventbuilder-20150320163004-lbnedaq2-37368.log

aggregator logfiles: 
lbnedaq3:/data/lbnedaq/daqlogs/aggregator/aggregator-20150320163004-lbnedaq2-37369.log
lbnedaq3:/data/lbnedaq/daqlogs/aggregator/aggregator-20150320163004-lbnedaq2-37367.log


The most significant things saved in this file for the end user of the DAQ system are the configuration chosen and the components chosen for the run (at the top of the file), and the wildcard for the pmt (ArtDAQ Process Management Tool) logfiles, which contain the output of the DAQ system, under "pmt logfile(s)" -- so, in this case, executing ls -ltr /data/lbnedaq/daqlogs/pmt/pmt-25771*.log would list the files containing the output of the DAQ in run 778. Concerning this last point, just be aware of the possibility that earlier DAQ sessions may have produced logfiles which also satisfy the wildcard, as the "25771" in the wildcard example provided refer to the process ID of the pmt process, which isn't necessarily unique for a given run.

The DAQInterface configuration file

DAQInterface uses some key information stored in a local configuration file, NOT to be confused with the run configuration described elsewhere on this wiki. Examples of this information are the hosts and ports (i.e., communication sockets) for artdaq processes, location of lbne-artdaq software, debug (log) level, etc). These can be changed while the DAQ is in its "stopped" state. The default DAQInterface configuration file is docs/config.txt (relative to /data/lbnedaq/daqarea/lbnerc); if you wish to change parameters, it's considered best practice not to edit this default file but rather to copy it to a file of the form "docs/config_<your username>.txt", and edit that file, passing it to the DAQInterface executable as described in the section on launching applications. It's expected that for normal operations, most parameters should not need to changed too often, and alterations will be primarily of developer or expert interest. If you ARE a developer/expert or are simply curious, take a look at the default DAQInterface configuration file; it should look something like the following:


lbne-artdaq: /data/lbnedaq/daqarea/lbne-artdaq-base/build_lbne-artdaq

PMT host: lbnedaq3
PMT port: 5400

pause before initialization: 5

# debug level can range from 0 to 3 (increasing order of verbosity)
debug level: 1

log directory:   /data/lbnedaq/daqlogs

record directory: /data/lbnedaq/run_records

# If this file is "config.txt" -- i.e., the standard, non-expert,
# non-testing version of the DAQInterface configuration file -- do not
# set "disable configuration check" to "true" 

disable configuration check: false

TDU XMLRPC port: 50008

EventBuilder host: lbnedaq2
EventBuilder port: 5235

EventBuilder host: lbnedaq2
EventBuilder port: 5236

Aggregator host: lbnedaq2
Aggregator port: 5265

Aggregator host: lbnedaq2
Aggregator port: 5266

The parameters you're most likely to change will be the following:

  • lbne-artdaq : the directory in which the desired build of lbne-artdaq to use is located; you would change this if you'd modified and built a personal copy of lbne-artdaq for hardware troubleshooting purposes, for example.
  • debug level : allows the user to set the verbosity level of the output to the screen; setting it to higher values creates greater verbosity, and as a practical matter, the range of settings is currently 0-3. Loosely speaking, "0" means minimal output (not much beyond simply announcing a transition is complete), "1" includes announcements of progress during transitions, "2" includes these announcements plus the values of certain variables as well as lbne-artdaq output, and "3" is primarily of developer interest.
  • disable configuration check : a boolean which can be set to "true" or "false"; normally this should be set to "false", which means that on the start transition, DAQInterface will intentionally fail and put itself into the "stopped" state if it discovers that edits have been made to the configuration directory since the last commit in that directory. If very frequent edits to the configuration directory are made during hardware debugging, however, switching this feature on to "true" can make life considerably easier.
  • TDU XMLRPC port : the value of the port through which DAQInterface will communicate with an XML-RPC server linked to the TDU. Needed for DAQInterface to send a sync pulse to the TDU at the beginning of the "start" and "resume" transitions. If this is set to 0 or a negative value, no sync pulse is attempted. For more on the TDU and its code, see Tom Dealtry's Wiki here

Adding new configurations to RunControl

Adding new DAQ Components to RunControl

Troubleshooting and FAQ

Common RunControl issues

How can I get a list of all RunControl commands?

lbnecmd help

will list all available commands for RunControl with some details for help.

No real configurations are listed by RunControl

If you run "lbnecmd listconfigs" and see something like:

Available configs (Name : description)

dummy : Dummy description

Instead of your expected configurations, RunControl is not able to connect to CfgMgr. Please make sure it's running; how to do this, described earlier in the wiki, is simply to execute "lbnecmd check", and launching the configuration manager, also described earlier, can be done via "launch_daq_applications.sh -c"

"lbnecmd check" returns "Unknown exception '<ProtocolError for localhost:50008/RPC2: -1 >'"

This sometimes happens when the TDU XML-RPC server has died. You can confirm this has happened by running "check_daq_applications.sh"; to relaunch it, run "launch_daq_applications.sh -t".

Common DAQInterface issues

As a preliminary: if something goes wrong when running the DAQ, there are two valuable sources of information:
  1. The pmt logfile(s), containing the output of the artdaq processes (including the output of the Art modules run within them -- so, for example, an exception throw from within a module would appear here). Where to find these logfiles is described in the section "Examining your output", earlier in this wiki.
  2. The RunControl logfile, /home/lbnedaq/.lbnerc.log . This contains both the information sent from DAQInterface to RunControl, as well as diagnostic logging (such as the traceback of exception throws within DAQInterface)

In general, when something goes wrong during the running of the DAQ -- e.g., an artdaq process dies, or a sync pulse is attempted when the TDU XML-RPC server isn't alive -- DAQInterface will impose upon itself the "recovery" transition, where it returns itself to the "stopped" state regardless of whatever state it may have currently been in. When this occurs, you'll see something like the following:

DAQInterface: "Recover" transition underway
JCF, 6/12/14 -- for now at least, "Recover" simply kills the artdaq processes

Recover transition complete; if running DAQInterface in the background, can press <enter> to return to shell prompt

If a recovery transition occurs, often you can begin using the DAQ again (starting with "lbnecmd init daq") without any issues. However, some issues may be persistent, or require manual intervention. A subset of those issues are described below. If any new ones come up, please contact John Freeman at .

On the initialize transition ("lbnecmd init daq"), you see "pmt.rb -p <port number> was already running on <hostname>"

An instance of artdaq's Process Management Tool program, tasked with creating the artdaq processes, is already running on the host and port you've requested for it in the DAQInterface configuration file passed to the DAQInterface application. This can happen if, for example, the DAQInterface program from a previous session was killed off before the usual "lbnecmd terminate daq" transition was issued, a transition which prompts DAQInterface to kill off the pmt.rb process it has control over. Generally, DAQInterface will clean this up and free the port in the process of returning itself to the "stopped" state, but in case it doesn't, there's a script called "kill_artdaq_processes.sh" which can be used in the following manner:

kill_artdaq_processes.sh <pmt host> <pmt port>

This script will both kill off the pmt.rb which is clogging the port, as well as the artdaq processes over which it has control.

On the initialize transition ("lbnecmd init daq"), you see "error: [Errno 111] Connection refused"

If a "Recover" is triggered and you can see via "lbnecmd check" that DAQInterface is in the "stopped" state, try initializing again. If that doesn't work, you can try increasing the value of the "pause before initialization" variable in the DAQInterface configuration file. Empirically, it appears there needs to be a pause of at least 5 seconds before the FHiCL documents can be successfully sent via XML-RPC to the processes; increasing this value may make it less likely that the "Connection refused" error occurs.

On the start transition ("lbnecmd start daq"), you see "TDU RESULT: socket.error caught: [Errno 111] Connection refused Is the XMLRPC server up?"

This error will trigger the Recover transition and return you to the stopped state. What it means is that there was an issue connecting to the TDU XML-RPC server (most likely, it wasn't running). Run "lbnecmd check" to see whether it's available; if not, launch it in the usual fashion (i.e., "launch_daq_applications.sh -t"), and then proceed with the standard DAQ transition sequence ("lbnecmd init daq", etc.)

Regardless of how high the debug level is set to, you don't see any output to screen when you issue transitions to DAQInterface

Chances are that DAQInterface was started in another terminal, and consequently, it's in that terminal where output will appear. If DAQInterface was started via the "launch_daq_applications.sh" script, then as of May-5-2015, all output is directed to /data/lbnedaq/daqlogs/daqinterface/DI.log, so running "tail -f /data/lbnedaq/daqlogs/daqinterface/DI.log" from another terminal will guarantee you can see DAQInterface output in real time. Please note this file is appended to, not overwritten, so only the end of the file will be of interest.

Common lbne-artdaq issues

The message "Failed to connect to shared memory segment" appears in the logfile

The full message is "Failed to connect to shared memory segment, errno = 22. Please check if a stale shared memory segment needs to be cleaned up. (ipcs, ipcrm -m <segId>)". Go to the system on which the process that issued the message is running (e.g., if it was an AggregatorMain process on lbnedaq2, go there) and run ipcs. You should see something like the following:

------ Shared Memory Segments --------
key        shmid      owner      perms      bytes      nattch     status      
0x40471518 4161536    lbnedaq    666        16777216   0          

This is the offending memory segment. To remove it, run the following (where here, we'll assume it's the one with shared memory ID 4161536, above-- of course you'll most likely see a different ID): ipcrm -m 4161536. Now, if you run ipcs again, the segment should not be listed; furthermore, the error should not appear again if you run again with the DAQ.