[Dsg-hallc_epics] HV backup not working

[Tyler Lemon]

Hi Steve,

After we talked yesterday about the problems you came across with the backup program trying to get the HV CSS screens integrated into Hall C's Makefile, I started working on some changes to the HV backup program that should fix most of the problem.

The main change is that the backup program has been modified to do away with using pyepics. It now uses only standard libraries and functions that are automatically included in Python and CSS installations. This makes it so there aren't any extra packages that would need to be installed to use the program.

In the CSS library, there was a function that gets the system path of the CSS workspace, allowing the file path dependencies to be removed.

I have also been able to remove the use of a channel reference text file noted yesterday and the hardcoded group names/numbers in you previous email by changing the program to get the record names from the .opi files rather than a text file or previous backup.

As for your previous email regarding the backup file containing only headers and not PVs, I'll look to see if that is the case with the new version and whether I can find the cause in the old version.

There's a few more things to clean up in it (putting better comments, making sure there's no redundant code), but when its done I will put the new version on cdaql2.

Best regards,
Tyler



________________________________
From: Stephen Wood <saw at jlab.org>
Sent: Wednesday, July 10, 2019 11:45 AM
To: Tyler Lemon
Subject: HV backup not working

Tyler:

In the copy of the scripts that I have under the cvxwrks account
(~cvxwrks/saw/EpicsHL/HV/CSS which comes from release5), backup is not
working.  I just get headers, but no PVs.  I traced it down to the line:

  if pv.split(':')[3] in props:

which never succeeds.

props is ['V0Setr','I0Setr','SVMaxr','RUpr','RDWnr']

and none of the PVs tested end in "r".  I could remove the r in the
props list, but technically it is the these readbacks that we want to
backup.


And for eventual improvement:  I notice that you explicitly list group
numbers and names in HV-bur.py.  (groups and systems lists).  Ideally
these should not be hardwired into the script, but derived from
configuration files.

I could arrange for my make file to put files like HMS.group, HMS.hvc,
SHMS.group and SHMS.hvc into the CSS directory and then you could search
for *.group and *.hvc to build up group and pv lists.  But I am open to
however you want to do this.

Thanks, Steve


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/dsg-hallc_epics/attachments/20190710/4277727e/attachment.html>