Re: seq 2.2.4 problem?

[Benjamin Franksen <benjamin.franksen@helmholtz-berlin.de>]

On 31.01.2017 20:26, Mark Rivers wrote:
> We upgraded our beamlines from base 3.14.12.5 and seq 2.2.3 to base
> 3.15.5 and seq 2.2.4.
> 
> This morning we observed a problem with Channel Access Security from
> ca_array_put from an SNL program which we had not seen before, and
> are wondering if it could be related to a change in the version of
> seq.
> 
> Configuration:
> 
> 
> -          A soft IOC (linux-x86_64) runs an SNL program that
> controls an undulator at the APS.
> 
> -          The APS undulator is behind a PV Gateway, and implements
> Channel Access Security.  Only certain computers and accounts are
> allowed to control the undulator, and they are only allowed to
> control it in user-beam mode, not during studies modes.
> 
> -          We started our the IOC when the APS was in studies mode,
> so the SNL program was not allowed to write to the undulator.
> 
> -          This morning the APS changed to user mode, so the SNL
> program should have been able to control the undulator.  However, it
> could not.  We had these errors in our log:
> 
> Mon Jan 30 17:58:17 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Mon Jan 30 17:59:00 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Mon Jan 30 17:59:00 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Mon Jan 30 17:59:00 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Mon Jan 30 17:59:00 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:21:58 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:21:58 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:22:01 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:22:01 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:23:12 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:23:12 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> Tue Jan 31 09:23:12 2017sevr=minor ca_array_put(typeToCA(type), count, var->chid, value): Write access denied
> 
> Note that the errors from Jan. 30 (yesterday) are expected because
> the APS was in studies mode. But at 08:00 on Jan. 31 (this morning) the APS
> switched to user mode and we should have been able to write to the
> undulator. However, we could not.
> 
> Rebooting our IOC, i.e. restarting the SNL program fixed the problem.
> 
> Can you think of any reason why seq 2.2.4 might be causing this
> issue? I thought I would ask before I begin to dig deeper and set up test IOCs.

What I can tell you is that the sequencer does not install an
access_rights_handler for its channels. IIUC this should not prevent it
from writing to the PVs once write access has been restored. Experts who
know more about the internals of the CA client library may chime in here
and correct me...

Cheers
Ben
-- 
"Make it so they have to reboot after every typo." ― Scott Adams

Attachment: signature.asc
Description: OpenPGP digital signature