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 | 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 |
<== Date ==> | <== Thread ==> |
---|
Subject: | Fix released (Re: CAUTION: Problem with areaDetector ImageJ PVA viewer) |
From: | Ralph Lange <[email protected]> |
To: | EPICS Tech Talk <[email protected]> |
Date: | Thu, 4 Oct 2018 09:13:44 +0200 |
Folks,
I have found a serious issue with the EPICS V4 Java client code which is exposed with the areaDetector ImageJ PVA viewer (EPICS_NTNDA_Viewer.java).
When the areaDetector IOC is not running the ImageJ plugin keeps trying to connect. I have found that the broadcast search requests have an ever-increasing number of duplicate PVs in them. The search requests normally only occur at 0.5 Hz or so. However, as the number of duplicate PVs per search increases, each search requires multiple Ethernet packets. After a few hours the number of broadcast packets is very large. I have seen those broadcasts consume >90% of the CPU time on all of my VME IOCs.
It appears that the problem is not in the ImageJ plugin, but rather in the underlying epics-pvaclient Java code. Marty Kraimer and I are planning to test a fix to that soon.
In the meantime I would discourage people from using the ImageJ EPICS_NTNDA_Viewer, unless you are sure to kill ImageJ when the areaDetector IOC is not running. Other applications using the epics-pvaclient Java library may also have this problem.
Mark