[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20050509: 20050507: nex2gini seg faults in Solaris 9/10



Robert,

The behavior of the WS F77 compiler with the -O is not outside
of the expected behavior for agressive optimization since it
tends to not allow input and output pointers to be the same-
which happens in GEMPAK. Some testing as to the different levels of
-On versus the default -O level might be useful, but not necessarily
going to be the same between releases.

I've had RADFRQ set to 1 and running non stop on X86 since this morning without
any problems, so I'm guessing that it is the radar files on your end causing 
problems.
Are you using -close on your FILE actions for the nexrad products? Are any of 
your 
nexrad files getting written/overwritten by receiveing the product more than 
once?

If you are able to isolate a file, we can look at it for symptoms. 

Steve Chiswell
Unidata User Support




>From: "Robert Mullenax" <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200505091857.j49IvFP3027985

>This is a multi-part message in MIME format.
>
>------_=_NextPart_001_01C554C8.E71CE2A2
>Content-Type: text/plain;
>       charset="iso-8859-1"
>Content-Transfer-Encoding: quoted-printable
>
>Steve,
>
>It doesn't produce a core file.  I also saw the issues with using -O and =
>WS10.  I compiled using WS9, and I tried your binary.
>
>Do you think it is a waste of my time to file a bug with Sun about -O =
>breaking GEMPAK?  What I saw was that the GEMPAK mapping stopped =
>working.
>
>I will try setting RADTIM instead of using current.
>
>Thanks,
>Robert
>
>
>=20
>
>-----Original Message-----
>From: Unidata Support
>To: Robert Mullenax
>Cc: 'address@hidden'
>Sent: 5/9/2005 1:30 PM
>Subject: 20050507: nex2gini seg faults in Solaris 9/10=20
>
>
>Robert,
>
>Can you provide any gdb output from a core file? I need something since
>I can't duplicate any problem here.
>
>I run here under X86 5.9 with WS10 compilers with no trouble. I did
>verify=20
>that WS10 again must not use the -O flag for the f77 compiler, so make
>sure=20
>you don't have that option set for your build. Is the crash duplicatable
>when=20
>processing the same SITE when you set the RADTIM time to a fixed time
>instead=20
>of current? If so, could be a zlib problem when reading the compressed
>product....
>especially if you have a site where the file has been appended. That is,
>I
>mean that the same station ID/time appears in the runtime list when the
>crash occurs.
>
>Steve Chiswell
>Unidata User Support
>
>
>
>
>
>>From: Robert Mullenax <address@hidden>
>>Organization: UCAR/Unidata
>>Keywords: 200505071725.j47HPWP3016079
>
>>The Unidata binary of GEMPAK 5.8.1 of nex2gini seg faults about 95% of
>the
>>time whne running under Solaris 9 and Solaris 10 x86.  A few times it
>will
>>work, the rest of the time it seg faults without producing a core file.
>>
>>I have also tried compiling locally and that does not help.  Setting
>system
>>statcksize larger does not help and neither does reducing kxky.
>>
>>Thanks,
>>Robert Mullenax
>>
>--
>************************************************************************
>Unidata User Support                                    UCAR Unidata
>(303)497-8643                                                  P.O. Box
>address@hidden                                   Boulder, CO
>------------------------------------------------------------------------
>Unidata WWW Service
>------------------------------------------------------------------------
>NOTE: All email exchanges with Unidata User Support are recorded in the
>Unidata inquiry tracking system and then made publicly available
>through the web.  If you do not want to have your interactions made
>available in this way, you must let us know in each email you send to
>us.
>
>------_=_NextPart_001_01C554C8.E71CE2A2
>Content-Type: text/html;
>       charset="iso-8859-1"
>Content-Transfer-Encoding: quoted-printable
>
><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
><HTML>
><HEAD>
><META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
>charset=3Diso-8859-1">
><META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
>6.5.7226.0">
><TITLE>RE: 20050507: nex2gini seg faults in Solaris 9/10 </TITLE>
></HEAD>
><BODY>
><!-- Converted from text/plain format -->
>
><P><FONT SIZE=3D2>Steve,<BR>
><BR>
>It doesn't produce a core file.&nbsp; I also saw the issues with using =
>-O and WS10.&nbsp; I compiled using WS9, and I tried your binary.<BR>
><BR>
>Do you think it is a waste of my time to file a bug with Sun about -O =
>breaking GEMPAK?&nbsp; What I saw was that the GEMPAK mapping stopped =
>working.<BR>
><BR>
>I will try setting RADTIM instead of using current.<BR>
><BR>
>Thanks,<BR>
>Robert<BR>
><BR>
><BR>
><BR>
><BR>
>-----Original Message-----<BR>
>From: Unidata Support<BR>
>To: Robert Mullenax<BR>
>Cc: 'address@hidden'<BR>
>Sent: 5/9/2005 1:30 PM<BR>
>Subject: 20050507: nex2gini seg faults in Solaris 9/10<BR>
><BR>
><BR>
>Robert,<BR>
><BR>
>Can you provide any gdb output from a core file? I need something =
>since<BR>
>I can't duplicate any problem here.<BR>
><BR>
>I run here under X86 5.9 with WS10 compilers with no trouble. I did<BR>
>verify<BR>
>that WS10 again must not use the -O flag for the f77 compiler, so =
>make<BR>
>sure<BR>
>you don't have that option set for your build. Is the crash =
>duplicatable<BR>
>when<BR>
>processing the same SITE when you set the RADTIM time to a fixed =
>time<BR>
>instead<BR>
>of current? If so, could be a zlib problem when reading the =
>compressed<BR>
>product....<BR>
>especially if you have a site where the file has been appended. That =
>is,<BR>
>I<BR>
>mean that the same station ID/time appears in the runtime list when =
>the<BR>
>crash occurs.<BR>
><BR>
>Steve Chiswell<BR>
>Unidata User Support<BR>
><BR>
><BR>
><BR>
><BR>
><BR>
>&gt;From: Robert Mullenax &lt;address@hidden&gt;<BR>
>&gt;Organization: UCAR/Unidata<BR>
>&gt;Keywords: 200505071725.j47HPWP3016079<BR>
><BR>
>&gt;The Unidata binary of GEMPAK 5.8.1 of nex2gini seg faults about 95% =
>of<BR>
>the<BR>
>&gt;time whne running under Solaris 9 and Solaris 10 x86.&nbsp; A few =
>times it<BR>
>will<BR>
>&gt;work, the rest of the time it seg faults without producing a core =
>file.<BR>
>&gt;<BR>
>&gt;I have also tried compiling locally and that does not help.&nbsp; =
>Setting<BR>
>system<BR>
>&gt;statcksize larger does not help and neither does reducing kxky.<BR>
>&gt;<BR>
>&gt;Thanks,<BR>
>&gt;Robert Mullenax<BR>
>&gt;<BR>
>--<BR>
>************************************************************************<=
>BR>
>**** &lt;<BR>
>Unidata User =
>Support&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=
>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&=
>nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
>UCAR Unidata<BR>
>Program &lt;<BR>
>(303)497-8643&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=
>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&=
>nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
>bsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
>sp;&nbsp;&nbsp; P.O. Box<BR>
>3000 &lt;<BR>
>address@hidden&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&=
>nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&n=
>bsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
>sp;&nbsp; Boulder, CO<BR>
>80307 &lt;<BR>
>------------------------------------------------------------------------<=
>BR>
>---- &lt;<BR>
>Unidata WWW Service<BR>
><A =
>HREF=3D"http://my.unidata.ucar.edu/content/support";>http://my.unidata.uca=
>r.edu/content/support</A>&nbsp; &lt;<BR>
>------------------------------------------------------------------------<=
>BR>
>---- &lt;<BR>
>NOTE: All email exchanges with Unidata User Support are recorded in =
>the<BR>
>Unidata inquiry tracking system and then made publicly available<BR>
>through the web.&nbsp; If you do not want to have your interactions =
>made<BR>
>available in this way, you must let us know in each email you send =
>to<BR>
>us.<BR>
></FONT>
></P>
>
></BODY>
></HTML>
>------_=_NextPart_001_01C554C8.E71CE2A2--
>
--
NOTE: All email exchanges with Unidata User Support are recorded in the
Unidata inquiry tracking system and then made publicly available
through the web.  If you do not want to have your interactions made
available in this way, you must let us know in each email you send to us.