Re: github ci problem: CAS: forcing disconnect from 127.0.0.1:57020

[Ralph Lange via Core-talk <core-talk at aps.anl.gov>]

Might be https://github.com/epics-base/epics-base/issues/943

On Wed, 26 Aug 2026 at 14:17, Simon Rose via Core-talk <core-talk at aps.anl.gov> wrote:

Hi Torsten -

Yes, I think I see the same thing. We actually saw it last week during the training at ESS but blamed it on network issues.

This can  be replicated on the 7.0 branch of epics-base quite easily: start an IOC and try (for example) to use caput to write to any record. In the IOC shell I see

CAS: forcing disconnect from 172.18.22.141:59117

And from the terminal trying to write to the IOC with caput I see

$ ./caput record_name 1     

Old : record_name                    0

CA.Client.Exception...............................................

    Warning: "Virtual circuit disconnect"

    Context: "172.18.22.141:5064"

    Source File: ../cac.cpp line 1241

    Current Time: Wed Aug 26 2026 14:15:06.470965000

..................................................................

CA.Client.Exception...............................................

    Warning: "Virtual circuit disconnect"

    Context: "op=0, channel=record_name, type=DBR_TIME_DOUBLE, count=1, ctx="172.18.22.141:5064""

    Source File: ../getCopy.cpp line 91

    Current Time: Wed Aug 26 2026 14:15:06.471381000

..................................................................

CA.Client.Exception...............................................

    Warning: "Identical process variable names on multiple servers"

    Context: "Channel: "record_name", Connecting to: 172.18.22.141:5064, Ignored: 192.168.205.1:5064"

    Source File: ../cac.cpp line 1324

    Current Time: Wed Aug 26 2026 14:15:06.475347000

..................................................................

Read operation timed out: PV data was not read.

New : record_name 

Something is wrong here... 

Cheers,

Simon

From: Core-talk <core-talk-bounces at aps.anl.gov> on behalf of Torsten Bögershausen via Core-talk <core-talk at aps.anl.gov>
Date: Wednesday, 26 August 2026 at 13:59
To: core-talk at aps.anl.gov <core-talk at aps.anl.gov>
Subject: github ci problem: CAS: forcing disconnect from 127.0.0.1:57020

Hej EPICS-experts,

since some time (July 10 2026 or 2026-07-10) we have a failure
on one of our github runners.

If someone wants to look, here is a good run:
<https://urldefense.us/v3/__https://github.com/EuropeanSpallationSource/m-epics-ethercatmc/actions/runs/29092838578__;!!G_uCfscf7eWS!Zp_A98G8tvAmH3xmEtRxKFl_0Ydm7Wz0MSRmQ1ZrULhuV4JUEssosb0rF6Cv3MthfRkOlH0LmK79aHPdMb8$ >

One of the failing runs is here:
<https://urldefense.us/v3/__https://github.com/EuropeanSpallationSource/m-epics-ethercatmc/actions/runs/32959143825/job/98147389788__;!!G_uCfscf7eWS!Zp_A98G8tvAmH3xmEtRxKFl_0Ydm7Wz0MSRmQ1ZrULhuV4JUEssosb0rF6Cv3MthfRkOlH0LmK79fn7GMoM$ >


Of course the tests run fine here on a local machine
(with probably sighly different versions)

So digging into the log file shows one problem:

Wed Aug 26 11:21:03 UTC 2026
+ ./runTests.sh ca://IOC:m1 100_Record-HOMF.py
CAS: forcing disconnect from 127.0.0.1:48202
#... and later, PYT stands for the output of pytest:
PYT:2026-08-26 11:21:04 AxisCom get ca://IOC:m1.VAL ret=100.11
PYT:2026-08-26 11:21:04 AxisCom get ca://IOC:m1.DMOV
PYT:cannot connect to IOC:m1.DMOV
PYT:2026-08-26 11:21:09 AxisCom get ca://IOC:m1.DMOV ret=None

Before I try to set up a system with the hopefully exact versions as
on github:
Does a bell ring for anybody here ?
The commits in EPICS base show that some additional checks had been
introduced when processing channel access messages in rsrv
However, we are using pyepics inside the test cases, which uses
the libca from base, so things should work.
(And yes, Timo, we are using mostly pvacess for 99% of the tests)

Any ideas are welcome
Best Regards
/Torsten