EPICS Controls Argonne National Laboratory

Experimental Physics and
Industrial Control System

1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  <20052006  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  <20052006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
<== Date ==> <== Thread ==>

Subject: RE: UDP send, errno=0x37
From: "Jeff Hill" <[email protected]>
To: "'Allison, Stephanie'" <[email protected]>, "'Geoff Savage'" <[email protected]>, "'EPICS Tech-Talk'" <[email protected]>
Date: Tue, 5 Jul 2005 13:52:46 -0600
> Finally, I discovered that the
> network port for the CPU somehow got reconfigured to 100/full (long
story).
> Going to 100/half solved the problem.  Hopefully, your problem will be
> solved by just increasing buffers.

As I recall, if the vxWorks IOC powers up when the switch is off it sets
full/half duplex to a default value, and appears to not re-auto-negotiate
full/half duplex again when the switch *is* powered up. The switch and the
IOC can end up inconsistent, and soft reboots do not help. Resetting the IOC
after the switch powers on appears to be the only solution. 

Very low throughput communication is allowed to proceed when the
configuration of full/half duplex is inconsistent. This can cause increased
pool consumption in the IOC's IP kernel due to backlog for messages pending
transmit. Use of casw has been useful when detecting these problems.

Jeff

> -----Original Message-----
> From: Allison, Stephanie [mailto:[email protected]]
> Sent: Tuesday, July 05, 2005 9:47 AM
> To: Geoff Savage; EPICS Tech-Talk
> Subject: RE: UDP send, errno=0x37
> 
> Hi Geoff,
> 
> > Over the weekend we received these error messages from an
> > mv5500 running vxworks 5.5 and epics 3.14.6.  As far as I can
> > tell this is an ENOBUFS error.  Is this the correct interpretation?
> 
> Yes.   Looking at target/h/errno.h in my vxWorks distribution, I see that
> 0x37 (55) is ENOBUFS.  I got similar messages on my vxWorks 5.4.2 target
> earlier this year.  I tried increasing buffers... I tried John Sinclair's
> vxworks patch that he provided here...:
> http://www.aps.anl.gov/epics/tech-talk/2004/msg00221.php
> I swapped CPU boards...  Nothing worked.  Finally, I discovered that the
> network port for the CPU somehow got reconfigured to 100/full (long
story).
> Going to 100/half solved the problem.  Hopefully, your problem will be
> solved by just increasing buffers.
> 
> 
> Stephanie Allison



References:
RE: UDP send, errno=0x37 Allison, Stephanie

Navigate by Date:
Prev: RE: Asyn driver for FLUKE 8508A Mark Rivers
Next: RE: EPICS 3.14.7 static client build Jeff Hill
Index: 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  <20052006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
Navigate by Thread:
Prev: RE: UDP send, errno=0x37 Allison, Stephanie
Next: EPICS 3.14.7 static client build Chris Slominski
Index: 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  <20052006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
ANJ, 02 Sep 2010 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·