On Mon, 18 Nov 2002, Neil R. Smith wrote:
Hi Robb,
I sent you the log files from interactive pqsurf as requested.
Did you receive it? Does it provide any clues? Do I need
to provide more info? Thanks, -Neil
Robb Kambic wrote:
>
> On Thu, 14 Nov 2002, Unidata Support wrote:
>
> >
> > ------- Forwarded Message
> >
> > >To: support@xxxxxxxxxxxxxxxx
> > >From: "Neil R. Smith" <neils@xxxxxxxxxxxxxxxxxx>
> > >Subject: pqsurf crashing in ldm 5.2
> > >Organization: UCAR/Unidata
> > >Keywords: 200211142352.gAENq6L05743
> >
> > Our pqsurf is crashing in our recently upgraded ldm 5.2 (from
> > 5.0.6). We are useing the same pqsurf.conf file.
> >
> > Here are the ldmd.conf errors:
> > Nov 14 23:02:01 5Q:coriolis rpc.ldmd[188942]: Starting Up (built: Nov 13
> > 2002 21:54:31)
> > Nov 14 23:02:01 5Q:coriolis pqbinstats[162465]: Starting Up (188942)
> > Nov 14 23:02:01 5Q:coriolis pqexpire[187333]: Starting Up
> > Nov 14 23:02:01 5Q:coriolis pqact[181011]: Starting Up
> > Nov 14 23:02:01 5Q:coriolis pqsurf[169954]: Starting Up (188942)
> > Nov 14 23:02:01 5Q:coriolis pqact[187141]: Starting Up
> > .
>
> Neil,
>
> A couple things to do is to make a few startup changes:
>
> -Comment out the pqexpire line with a "#" in the etc/ldmd.conf file. That
> process is not needed any more and it might be causing some problems.
> - remake the ldmqueue and pqsurfqueue. ie
>
> % ldmadmin stop
> % ldmadmin delqueue
> % ldmadmin mkqueue
> % ldmadmin delsurfqueue
> % ldmadmin mlsurfqueue
> % ldmadmin start
>
> > . (no intervening proc.#169954 messages)
> > .
> > Nov 14 23:03:33 3Q:coriolis pqsurf[169954]: surface_split: Can't handle
> > MESSAGE_TYPE_UNKNOWN
> > Nov 14 23:03:33 3Q:coriolis pqsurf[169954]: HDR + PIL: SXHI
> > Nov 14 23:03:34 3Q:coriolis pqsurf[169954]: surface_split: Can't handle
> > MESSAGE_TYPE_UNKNOWN
> > Nov 14 23:03:34 3Q:coriolis pqsurf[169954]: HDR + PIL: SXHI
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: child 187141 terminated by
> > signal 10
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: Exiting
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: Queue usage (bytes):
> > 72784
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: (nregions):
> > 377
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: Number of products 27
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: Number of observations 135
> > Nov 14 23:03:36 5Q:coriolis pqsurf[169954]: Number of dups 2
> >
>
> - another idea is to run the pqsurf in verbose mode from the command line.
>
> - comment out pqsurf in the etc/ldmd.conf file
> - run pqsurf from command line as user ldm
>
> % pqsurf -v -l -
>
> This send output to the terminal. If you replace -l - with -l log it will
> put the output in file log
>
>
> See comment below first
>
> > Here is the ldmd.conf entry:
> > exec "pqsurf -p ^S[APX] -Q /data/pqsurf.pq /unidata/ldm/etc/pqsurf.conf"
>
> I would change the above line to
>
> exec "pqsurf -p ^S(A....|P....|XUS8.|XUS91) -q /data/pqsurf.pq
> /unidata/ldm/etc/pqsurf.conf"
>
> You might be getting some radar products or some other products wrongly
> causing pqsurf to crash.
>
> If none of the above works, send me info about your machine and the pqsurf
> log file from the command line.
>
> Robb...
>
> >
> > Here are the pqsurf.conf entries:
> > WMO ^sao .. (...) ([0-3][0-9])([0-2][0-9])
> > DBFILE /data/ddplus/db_sao/(\2:yy)(\2:mm)\2\3.SA_DB \1
> > WMO ^sao ... (...) ([0-3][0-9])([0-2][0-9])
> > DBFILE /data/ddplus/db_sao/(\2:yy)(\2:mm)\2\3.SA_DB \1
> > WMO ^metar (....) ([0-3][0-9])([0-2][0-9])
> > DBFILE /data/ddplus/db_metar/(\2:yy)(\2:mm)\2\3.METAR_DB \1
> > WMO ^speci (....) ([0-3][0-9])([0-2][0-9])
> > DBFILE /data/ddplus/db_metar/(\2:yy)(\2:mm)\2\3.METAR_DB \1
> >
> > Is any of the above config. antiquated and causing the
> > problem?
Neil,
We had a power outage today, it's a pita to get back your environment etc.
Anyhow, I look at the logs today. It seems the problem isn't with pqsurf
per se, it's the pqact running on the pqsurf.conf that exited. Since
pqact usuallly doesn't fail unless a entry in the file causes a problem.
I notice all your entires are DBFILE ones. I wondering if you could
comment those out with a '#' and make a entry that writes to a file? This
way we could determine if it's the DBFILE entries or not. How about using
a entry like this:
WMO ^metar (....) ([0-3][0-9])([0-2][0-9])
file /data/ddplus/(\2:yy)(\2:mm)\2\3.wmo
This will file all the metars in a file like: /data/ddplus/02111812.wmo
Rememeber the white spaces needs to be tabs.
Let me know what you find out.
Robb...
> > Thanks, -Neil
> > --
> > Neil R. Smith, Comp. Sys. Mngr. neils@xxxxxxxxxxxxxxxxxx
> > Dept. Atmospheric Sci., Texas A&M Univ. 979/845-6272 FAX:979/862-4466
> >
> >
> > ------- End of Forwarded Message
> >
> >
>
> ==============================================================================
> Robb Kambic Unidata Program Center
> Software Engineer III Univ. Corp for Atmospheric Research
> rkambic@xxxxxxxxxxxxxxxx WWW: http://www.unidata.ucar.edu/
> ==============================================================================
--
Neil R. Smith, Comp. Sys. Mngr. neils@xxxxxxxxxxxxxxxxxx
Dept. Atmospheric Sci., Texas A&M Univ. 979/845-6272 FAX:979/862-4466
==============================================================================
Robb Kambic Unidata Program Center
Software Engineer III Univ. Corp for Atmospheric Research
rkambic@xxxxxxxxxxxxxxxx WWW: http://www.unidata.ucar.edu/
==============================================================================