[Halld-online] [New Logentry] DAQ Testing

[davidl at jlab.org]

Logentry Text:
--
mcROL with TI triggering:
Today we wish to try using MC data, but have triggers come from the TS using a 
pulser. In addition to testing this function, a goal is to form a mechanism that will
quickly crash one of the gluon computers so that we may more easily debug that
issue. To do this a new readout list was created “mcROL_ti.so” and a new
CODA configuration: hd_all-cdc-fdc_mc

This configuration is set to use direct connections everywhere (including from
the ROCs to the DCs). No multi-casting is configured.

The link from SEB to ER had the new “SingleEventOut” box checked. This 
was not checked for any of the other connections.

All non-ROC components are set to run on gluon51.

1st attempt:
  - Existing coda_roc and coda_ts processes on the FCAL crates and BCAL crates
    were killed so that they could be started from the hdops account
  - Configure suceeded
  - Download failed due to SEB0 with message:
    “*** cannot load transport class***”
  
2nd attempt:
  - Simply hit “Configure”. No processes restarted and nothing changed.
  - Configure succeeded
  - Download failed with same SEB0 error

3rd attempt:
  - Uncheck the “SingleEventOut” box in SEB to ER link
  - Stop and restart all processes (this may be needed for the relink script to be
    run again.)
    - Actually, this does seem to re-run it, but this does overwrite changes made to
      the RTV “CODA_ROL”. Reset this in rcgui to point to the new ROL location again.
  - Configure succeeded
  - Download failed with same “cannot load transport class” error in SEB

4th attempt:
  - Switch to hd_all_mc3 configuration
  - Configure succeeded
  - Download succeeded
  - Prestart succeeded
  - Go seemed to succeed, but only 386 events went through before it stopped.
    no errors were noted

- Sergey comes in and has some experience with the “cannot load transport class”
  error. We remove the SEB0 component from the hd_all-cdc-fdc_mc configuration
  and add it back and save as a new configuration: hd_all-cdc-fdc_mc.tmp
  - Configure succeeded
  - Download failed. The SEB did not have the transport class error, but all 3
    DC’s did
  - Compare files in COOL_HOME and find the new SEB0.xml has an et_hdops…
    while the old has et_furletov… The DC’s all still have the furletov name. This
    apparently was a fix that was made to use the current expid/session. Removing
    and re-adding the component apparently gives names consistent with what is
    expected elsewhere in the code. Try doing global replace of “furletov” with
    “hdops” via command line.
  - The global replace did not work. There must be other places that need to be
     changed. Go back to jcedit and delete DC and SEB components and re-add
     and then re-link all ROCs
  - This works. 

5th attempt:
  - Configure succeeded
  - Download succeeded
  - Prestart failed. Several components died

Update Afecs from repository

6th attempt: Starting processes from scratch
  - Using old (bad) configuration to show DAQ guys errors
  - SEB failed with “cannot load transport class” error
  - rcgui reports exception (jana.lang.UnsatisfiedLinkError) but still runs
    - some windows (data rate) not available. Possibly due to 

7th attempt: Starting again from scratch with new configuration where all
links were remade
  - Configure succeeded
  - Download failed (mcROL_ti.so link not in right directory)
  - Fixed and re-tried
  - Download failed for TSG because it failed to find the TS card.
  - Dave A. looked with another utility and could see the card. Suspect a ts library
    may be out of date. Try compiling ts_list.so again against latests libraries
    from Bryan.

8th attempt: Using recompiled ts_list.so
  - Configure succeeded
  - Download failed due to undefined symbol when loading new libts.so. Looks
    to be missing most recent libjvme.so. Copy over and recompile once more

9th attempt
  - Configure succeeded
  - Download succeeded
  - Prestart failed due to DC ET systems having the event size set too small.
    They were set to 256 bytes and needed to be 2100 bytes.

10th attempt: 
  - Configure succeeded
  - Download succeeded
  - Prestart succeeded
  - Go Failed in DC’s due to ROC data being in wrong format. This
    is due to mcROL_ti not properly opening a bank to write to. Modified
   and recompiled.

11th attempt:
  - Configure succeeded
  - Download succeeded
  - Prestart succeeded
  - Go succeeded, but only pre start and go events came through. 
  - Need to remove roctof3 since it is not a DAQ crate (discriminators only)

12th attempt:
  - Configure succeeded
  - Download succeeded
  - Prestart succeeded
  - Go succeeded, but again stopped at 2 events. This was due to the VME
    semaphore being left in a taken state when the coda_roc process was
    killed on the previous attempt. This has been addressed already in more
    recent versions of the tiprimary_list.c which will automatically clear the
    semaphore in pre start if it finds it hung. Update to the newest
    tiprimary_list.c and recompile

13th attempt:
  - Configure succeeded
  - Download succeeded
  - Prestart succeeded
  - Go succeeded. Events flowed at ~4Hz.
  - Ran for ~ 761 and then we ended the run
  - End Succeeded



---

This is a plain text email for clients that cannot display HTML.  The full logentry can be found online at https://logbooks.jlab.org/entry/3289736
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/halld-online/attachments/20140717/9dd282c5/attachment.html