| 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 <2022> 2023 2024 2025 | Index | 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 <2022> 2023 2024 2025 |
| <== Date ==> | <== Thread ==> |
|---|
| Subject: | Re: Communication error |
| From: | Matt Rippa via Tech-talk <tech-talk at aps.anl.gov> |
| To: | Michael Davidsaver <mdavidsaver at gmail.com> |
| Cc: | EPICS tech-talk <tech-talk at aps.anl.gov>, Francisco Ramos <framoslandaeta at gmail.com> |
| Date: | Tue, 15 Mar 2022 12:40:30 -1000 |
On 3/10/22 08:43, Francisco Ramos via Tech-talk wrote:
> ...
> I don't know if EPICS 7 should be compatible with version ( 3.13.9 ), if not, I would appreciate if you could guide us on how to fix it. Right now we are not planning to migrate the subsystems we have compiled with that version of EPICS.
The short answer is to recommend putting a cagateway built against Base <= 3.15
in between any <= 3.13 clients and >=3.14 servers.
Note that this is only needed in one direction. New (>=3.14) clients can still
connect to any CA server version.
The longer story goes back to 2017 and:
https://code.launchpad.net/~epics-core/epics-base/camodern/+merge/306371
Where I was removing support for CA protocol versions circa Base <=3.12 which
depend on handling of UDP broadcasts which these days seem like an invitation
for abuse.
However, it seems that this change also effected 3.13.x era protocol versions.
This wasn't noticed at the time, and I and others only came to suspect it
last year.
https://github.com/epics-base/epics-base/pull/141/files/29f9d4db197487f9d81a52e5b5f3235532990172#r602439829
Which is what prompted me to put at a call for information.
https://epics.anl.gov/tech-talk/2021/msg00484.php
The response wasn't overwhelming. I did get one sample, which wasn't by
itself conclusive for me. (cf. ca-search-3.13.pcapng.gz) Also, the site
which provided this sample wasn't reporting any problems. With no urgency,
I stopped looking into this.
https://github.com/mdavidsaver/cashark/tree/master/test
I guess one moral of this story is that none of the Base developers test with
Base < 3.14 anymore (I never have). I don't expect this to change, which makes
situations like this seem inevitable.