[Met_help] [rt.rap.ucar.edu #78586] History for Problem running point_stat with HRRR_AK GRIB2 output
John Halley Gotway via RT
met_help at ucar.edu
Fri Nov 4 13:40:45 MDT 2016
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi, (cc'ing a couple of people at ESRL/GSD)
I'm running MET 5.2 (provisioned in a Docker Container in late September -
now I see some patches have been issued on 10 October). But, I'm not sure
that my problem would be related to these.
When I run point_stat with the default config file, using a GRIB2 HRRR_AK
(from ESRL/GSD group) I get a seg fault:
$ point_stat testdata/1630800001200.wrfprs.gr2 testdata/obs.nc
pointstat.conf
DEBUG 1: Default Config File:
/home/wrfuser/met-5.2/share/met/config/PointStatConfig_default
DEBUG 1: User Config File: pointstat.conf
Segmentation fault (core dumped)
However, when I use "cnvgrid -g21 ....." to convert the HRRR-AK to GRIB1,
I'm able to run this successfully and get a reasonable MPR output file.
In looking at documentation, I see that I can't use the "obs = fcst;"
statement in the config file if I'm using a non-GRIB1 forecast file, so I
know I will need to play around with this. But, in the long run, should I
even expect to be able to run this with GRIB2 input now? According to your
FAQ at
http://www.dtcenter.org/met/users/support/faqs/index.php
GRIB2 is not supported, but I'm thinking that might be very outdated??
Thanks for any guidance,
Don
---
Don Morton, Owner/Manager
Boreal Scientific Computing LLC
Fairbanks, Alaska USA
http://www.borealscicomp.com/
http://www.borealscicomp.com/Miscellaneous/MortonBio/
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: John Halley Gotway
Time: Thu Nov 03 16:38:16 2016
Don,
Yes, GRIB2 inputs should work fine.
Boy you're right, those FAQ's are extremely outdated! That's the
trouble
with writing something down. You need to keep it updated. I'll make
a
note that we need to update the content there.
My initial guess is that this is related to the version of the grib2c
library to which MET is linking. I suspect that recompiling the
grib2c
library and then recompiling MET against that updated library would
fix
it. Try version 1.6.0 of g2clib:
http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
Please let me know if updating g2clib and recompiling does not fix it.
That's interesting that you're running MET through a docker container.
I'm
working this week on providing exactly that for the MET online
tutorial.
Is that MET docker container available somewhere? Can you point me to
it?
Thanks,
John Halley Gotway
On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT <met_help at ucar.edu>
wrote:
>
> Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
> Transaction: Ticket created by don.morton at borealscicomp.com
> Queue: met_help
> Subject: Problem running point_stat with HRRR_AK GRIB2 output
> Owner: Nobody
> Requestors: don.morton at borealscicomp.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
>
>
> Hi, (cc'ing a couple of people at ESRL/GSD)
>
> I'm running MET 5.2 (provisioned in a Docker Container in late
September -
> now I see some patches have been issued on 10 October). But, I'm
not sure
> that my problem would be related to these.
>
> When I run point_stat with the default config file, using a GRIB2
HRRR_AK
> (from ESRL/GSD group) I get a seg fault:
>
> $ point_stat testdata/1630800001200.wrfprs.gr2 testdata/obs.nc
> pointstat.conf
> DEBUG 1: Default Config File:
> /home/wrfuser/met-5.2/share/met/config/PointStatConfig_default
> DEBUG 1: User Config File: pointstat.conf
> Segmentation fault (core dumped)
>
> However, when I use "cnvgrid -g21 ....." to convert the HRRR-AK to
GRIB1,
> I'm able to run this successfully and get a reasonable MPR output
file.
>
> In looking at documentation, I see that I can't use the "obs =
fcst;"
> statement in the config file if I'm using a non-GRIB1 forecast file,
so I
> know I will need to play around with this. But, in the long run,
should I
> even expect to be able to run this with GRIB2 input now? According
to your
> FAQ at
>
> http://www.dtcenter.org/met/users/support/faqs/index.php
>
> GRIB2 is not supported, but I'm thinking that might be very
outdated??
>
> Thanks for any guidance,
>
> Don
>
> ---
> Don Morton, Owner/Manager
> Boreal Scientific Computing LLC
> Fairbanks, Alaska USA
> http://www.borealscicomp.com/
> http://www.borealscicomp.com/Miscellaneous/MortonBio/
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: Don Morton
Time: Thu Nov 03 18:04:05 2016
Hi John,
I just "invited" you (via johnhg at ucar.edu) to share my bitbucket repo
with
the Dockerfile and various files. Should be everything you need to
build
and run the image.
It's very experimental, as my long term goal is to build a "MET as a
service in AWS" for processing HRRRAK files. To create all the libs
and
MET, I fired up a simple container, mounted to one of my host
directories,
then did all the compiling/installing in there. The host directories
are
now part of the Docker provisioning system, so they get copied into
the
container.
Not sure I'm really happy with the approach, but I don't want to go
through
a full compilation each time I create a new container. Tradeoffs...
I
think ideally, one might create a MET package that could be installed
with
yum, but I haven't gotten that far.
I am, indeed, using g2clib 1.6.0 - followed all the instructions in
the MET
v5.2 documentation. But, thanks for the response - I'll poke around a
bit
more. I think I was also having problems with the NAM GRIB2 files...
Best,
Don
---
Don Morton, Owner/Manager
Boreal Scientific Computing LLC
Fairbanks, Alaska USA
http://www.borealscicomp.com/
http://www.borealscicomp.com/Miscellaneous/MortonBio/
On Thu, Nov 3, 2016 at 10:38 PM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:
> Don,
>
> Yes, GRIB2 inputs should work fine.
>
> Boy you're right, those FAQ's are extremely outdated! That's the
trouble
> with writing something down. You need to keep it updated. I'll
make a
> note that we need to update the content there.
>
> My initial guess is that this is related to the version of the
grib2c
> library to which MET is linking. I suspect that recompiling the
grib2c
> library and then recompiling MET against that updated library would
fix
> it. Try version 1.6.0 of g2clib:
> http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
>
> Please let me know if updating g2clib and recompiling does not fix
it.
>
> That's interesting that you're running MET through a docker
container. I'm
> working this week on providing exactly that for the MET online
tutorial.
> Is that MET docker container available somewhere? Can you point me
to it?
>
> Thanks,
> John Halley Gotway
>
>
> On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
> > Transaction: Ticket created by don.morton at borealscicomp.com
> > Queue: met_help
> > Subject: Problem running point_stat with HRRR_AK GRIB2 output
> > Owner: Nobody
> > Requestors: don.morton at borealscicomp.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
> >
> >
> > Hi, (cc'ing a couple of people at ESRL/GSD)
> >
> > I'm running MET 5.2 (provisioned in a Docker Container in late
September
> -
> > now I see some patches have been issued on 10 October). But, I'm
not
> sure
> > that my problem would be related to these.
> >
> > When I run point_stat with the default config file, using a GRIB2
HRRR_AK
> > (from ESRL/GSD group) I get a seg fault:
> >
> > $ point_stat testdata/1630800001200.wrfprs.gr2 testdata/obs.nc
> > pointstat.conf
> > DEBUG 1: Default Config File:
> > /home/wrfuser/met-5.2/share/met/config/PointStatConfig_default
> > DEBUG 1: User Config File: pointstat.conf
> > Segmentation fault (core dumped)
> >
> > However, when I use "cnvgrid -g21 ....." to convert the HRRR-AK to
GRIB1,
> > I'm able to run this successfully and get a reasonable MPR output
file.
> >
> > In looking at documentation, I see that I can't use the "obs =
fcst;"
> > statement in the config file if I'm using a non-GRIB1 forecast
file, so I
> > know I will need to play around with this. But, in the long run,
should
> I
> > even expect to be able to run this with GRIB2 input now?
According to
> your
> > FAQ at
> >
> > http://www.dtcenter.org/met/users/support/faqs/index.php
> >
> > GRIB2 is not supported, but I'm thinking that might be very
outdated??
> >
> > Thanks for any guidance,
> >
> > Don
> >
> > ---
> > Don Morton, Owner/Manager
> > Boreal Scientific Computing LLC
> > Fairbanks, Alaska USA
> > http://www.borealscicomp.com/
> > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> >
> >
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: John Halley Gotway
Time: Fri Nov 04 09:55:37 2016
Don,
Thanks for the bitbucket invite. Unfortunately, I get this error
message:
The owner of this repository (borealscicomp) is already at their
plan
limit and will need to upgrade before adding more users.
I'm still very new to container world, but here's the github repo with
version I'm working on:
https://github.com/NCAR/container-dtc-met
This is the dockerfile, not the pre-built image, but if you have
docker
installed, the directions in this readme should enable you to build
the
image:
https://github.com/NCAR/container-dtc-met/blob/master/met-
5.2/README.docker.txt
The goal of this container is to make MET easily available for those
who
want to go through the online tutorial exercises. So we have them
build
the image and then /bin/bash into the environment to go through the
tutorial.
However, I still don't have X11 forwarding working... so displaying
GUI's
in the container doesn't currently work, such as ncview and gv. Have
you
run into similar issues? Any suggestions?
As for the GRIB2-C library, I was able to install that as a package
using
yum for CentOS:
yum -y install g2clib-devel
I checked that MET is able to read GRIB2 files in the container with
no
segfault. If you're compiling the GRIB2-C library from source, you
could
try that package instead to see if you have better luck.
John
On Thu, Nov 3, 2016 at 6:04 PM, Don Morton via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
>
> Hi John,
>
> I just "invited" you (via johnhg at ucar.edu) to share my bitbucket
repo with
> the Dockerfile and various files. Should be everything you need to
build
> and run the image.
>
> It's very experimental, as my long term goal is to build a "MET as a
> service in AWS" for processing HRRRAK files. To create all the libs
and
> MET, I fired up a simple container, mounted to one of my host
directories,
> then did all the compiling/installing in there. The host
directories are
> now part of the Docker provisioning system, so they get copied into
the
> container.
>
> Not sure I'm really happy with the approach, but I don't want to go
through
> a full compilation each time I create a new container. Tradeoffs...
I
> think ideally, one might create a MET package that could be
installed with
> yum, but I haven't gotten that far.
>
> I am, indeed, using g2clib 1.6.0 - followed all the instructions in
the MET
> v5.2 documentation. But, thanks for the response - I'll poke around
a bit
> more. I think I was also having problems with the NAM GRIB2
files...
>
> Best,
>
> Don
>
> ---
> Don Morton, Owner/Manager
> Boreal Scientific Computing LLC
> Fairbanks, Alaska USA
> http://www.borealscicomp.com/
> http://www.borealscicomp.com/Miscellaneous/MortonBio/
>
> On Thu, Nov 3, 2016 at 10:38 PM, John Halley Gotway via RT <
> met_help at ucar.edu> wrote:
>
> > Don,
> >
> > Yes, GRIB2 inputs should work fine.
> >
> > Boy you're right, those FAQ's are extremely outdated! That's the
trouble
> > with writing something down. You need to keep it updated. I'll
make a
> > note that we need to update the content there.
> >
> > My initial guess is that this is related to the version of the
grib2c
> > library to which MET is linking. I suspect that recompiling the
grib2c
> > library and then recompiling MET against that updated library
would fix
> > it. Try version 1.6.0 of g2clib:
> > http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
> >
> > Please let me know if updating g2clib and recompiling does not fix
it.
> >
> > That's interesting that you're running MET through a docker
container.
> I'm
> > working this week on providing exactly that for the MET online
tutorial.
> > Is that MET docker container available somewhere? Can you point
me to
> it?
> >
> > Thanks,
> > John Halley Gotway
> >
> >
> > On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
> > > Transaction: Ticket created by don.morton at borealscicomp.com
> > > Queue: met_help
> > > Subject: Problem running point_stat with HRRR_AK GRIB2
output
> > > Owner: Nobody
> > > Requestors: don.morton at borealscicomp.com
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586
> >
> > >
> > >
> > > Hi, (cc'ing a couple of people at ESRL/GSD)
> > >
> > > I'm running MET 5.2 (provisioned in a Docker Container in late
> September
> > -
> > > now I see some patches have been issued on 10 October). But,
I'm not
> > sure
> > > that my problem would be related to these.
> > >
> > > When I run point_stat with the default config file, using a
GRIB2
> HRRR_AK
> > > (from ESRL/GSD group) I get a seg fault:
> > >
> > > $ point_stat testdata/1630800001200.wrfprs.gr2 testdata/obs.nc
> > > pointstat.conf
> > > DEBUG 1: Default Config File:
> > > /home/wrfuser/met-5.2/share/met/config/PointStatConfig_default
> > > DEBUG 1: User Config File: pointstat.conf
> > > Segmentation fault (core dumped)
> > >
> > > However, when I use "cnvgrid -g21 ....." to convert the HRRR-AK
to
> GRIB1,
> > > I'm able to run this successfully and get a reasonable MPR
output file.
> > >
> > > In looking at documentation, I see that I can't use the "obs =
fcst;"
> > > statement in the config file if I'm using a non-GRIB1 forecast
file,
> so I
> > > know I will need to play around with this. But, in the long
run,
> should
> > I
> > > even expect to be able to run this with GRIB2 input now?
According to
> > your
> > > FAQ at
> > >
> > > http://www.dtcenter.org/met/users/support/faqs/index.php
> > >
> > > GRIB2 is not supported, but I'm thinking that might be very
outdated??
> > >
> > > Thanks for any guidance,
> > >
> > > Don
> > >
> > > ---
> > > Don Morton, Owner/Manager
> > > Boreal Scientific Computing LLC
> > > Fairbanks, Alaska USA
> > > http://www.borealscicomp.com/
> > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> > >
> > >
> >
> >
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: Don Morton
Time: Fri Nov 04 10:16:48 2016
Hi John,
I guess I have a five-user limit (including myself), so just removed
two of
them who don't need access. I'll resend you an invite in a minute.
My container is also CentOS (7), but I complied g2clib from scratch,
so
I'll try it your way, too. If that fails, maybe I'll backtrack to Met
5.1,
though hate to do that.
I'll look at the X stuff later. Seems like something I've run into,
but
can't remember the resolution now.
Best,
Don
---
Don Morton, Owner/Manager
Boreal Scientific Computing LLC
Fairbanks, Alaska USA
http://www.borealscicomp.com/
http://www.borealscicomp.com/Miscellaneous/MortonBio/
On Fri, Nov 4, 2016 at 3:55 PM, John Halley Gotway via RT
<met_help at ucar.edu
> wrote:
> Don,
>
> Thanks for the bitbucket invite. Unfortunately, I get this error
message:
> The owner of this repository (borealscicomp) is already at their
plan
> limit and will need to upgrade before adding more users.
>
> I'm still very new to container world, but here's the github repo
with
> version I'm working on:
> https://github.com/NCAR/container-dtc-met
>
> This is the dockerfile, not the pre-built image, but if you have
docker
> installed, the directions in this readme should enable you to build
the
> image:
>
> https://github.com/NCAR/container-dtc-met/blob/master/
> met-5.2/README.docker.txt
>
> The goal of this container is to make MET easily available for those
who
> want to go through the online tutorial exercises. So we have them
build
> the image and then /bin/bash into the environment to go through the
> tutorial.
>
> However, I still don't have X11 forwarding working... so displaying
GUI's
> in the container doesn't currently work, such as ncview and gv.
Have you
> run into similar issues? Any suggestions?
>
> As for the GRIB2-C library, I was able to install that as a package
using
> yum for CentOS:
> yum -y install g2clib-devel
>
> I checked that MET is able to read GRIB2 files in the container with
no
> segfault. If you're compiling the GRIB2-C library from source, you
could
> try that package instead to see if you have better luck.
>
> John
>
> On Thu, Nov 3, 2016 at 6:04 PM, Don Morton via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
> >
> > Hi John,
> >
> > I just "invited" you (via johnhg at ucar.edu) to share my bitbucket
repo
> with
> > the Dockerfile and various files. Should be everything you need
to build
> > and run the image.
> >
> > It's very experimental, as my long term goal is to build a "MET as
a
> > service in AWS" for processing HRRRAK files. To create all the
libs and
> > MET, I fired up a simple container, mounted to one of my host
> directories,
> > then did all the compiling/installing in there. The host
directories are
> > now part of the Docker provisioning system, so they get copied
into the
> > container.
> >
> > Not sure I'm really happy with the approach, but I don't want to
go
> through
> > a full compilation each time I create a new container.
Tradeoffs... I
> > think ideally, one might create a MET package that could be
installed
> with
> > yum, but I haven't gotten that far.
> >
> > I am, indeed, using g2clib 1.6.0 - followed all the instructions
in the
> MET
> > v5.2 documentation. But, thanks for the response - I'll poke
around a
> bit
> > more. I think I was also having problems with the NAM GRIB2
files...
> >
> > Best,
> >
> > Don
> >
> > ---
> > Don Morton, Owner/Manager
> > Boreal Scientific Computing LLC
> > Fairbanks, Alaska USA
> > http://www.borealscicomp.com/
> > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> >
> > On Thu, Nov 3, 2016 at 10:38 PM, John Halley Gotway via RT <
> > met_help at ucar.edu> wrote:
> >
> > > Don,
> > >
> > > Yes, GRIB2 inputs should work fine.
> > >
> > > Boy you're right, those FAQ's are extremely outdated! That's
the
> trouble
> > > with writing something down. You need to keep it updated. I'll
make a
> > > note that we need to update the content there.
> > >
> > > My initial guess is that this is related to the version of the
grib2c
> > > library to which MET is linking. I suspect that recompiling the
grib2c
> > > library and then recompiling MET against that updated library
would fix
> > > it. Try version 1.6.0 of g2clib:
> > > http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
> > >
> > > Please let me know if updating g2clib and recompiling does not
fix it.
> > >
> > > That's interesting that you're running MET through a docker
container.
> > I'm
> > > working this week on providing exactly that for the MET online
> tutorial.
> > > Is that MET docker container available somewhere? Can you point
me to
> > it?
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > > On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
> > > > Transaction: Ticket created by don.morton at borealscicomp.com
> > > > Queue: met_help
> > > > Subject: Problem running point_stat with HRRR_AK GRIB2
output
> > > > Owner: Nobody
> > > > Requestors: don.morton at borealscicomp.com
> > > > Status: new
> > > > Ticket <URL: https://rt.rap.ucar.edu/rt/
> Ticket/Display.html?id=78586
> > >
> > > >
> > > >
> > > > Hi, (cc'ing a couple of people at ESRL/GSD)
> > > >
> > > > I'm running MET 5.2 (provisioned in a Docker Container in late
> > September
> > > -
> > > > now I see some patches have been issued on 10 October). But,
I'm not
> > > sure
> > > > that my problem would be related to these.
> > > >
> > > > When I run point_stat with the default config file, using a
GRIB2
> > HRRR_AK
> > > > (from ESRL/GSD group) I get a seg fault:
> > > >
> > > > $ point_stat testdata/1630800001200.wrfprs.gr2 testdata/obs.nc
> > > > pointstat.conf
> > > > DEBUG 1: Default Config File:
> > > > /home/wrfuser/met-5.2/share/met/config/PointStatConfig_default
> > > > DEBUG 1: User Config File: pointstat.conf
> > > > Segmentation fault (core dumped)
> > > >
> > > > However, when I use "cnvgrid -g21 ....." to convert the HRRR-
AK to
> > GRIB1,
> > > > I'm able to run this successfully and get a reasonable MPR
output
> file.
> > > >
> > > > In looking at documentation, I see that I can't use the "obs =
fcst;"
> > > > statement in the config file if I'm using a non-GRIB1 forecast
file,
> > so I
> > > > know I will need to play around with this. But, in the long
run,
> > should
> > > I
> > > > even expect to be able to run this with GRIB2 input now?
According
> to
> > > your
> > > > FAQ at
> > > >
> > > > http://www.dtcenter.org/met/users/support/faqs/index.php
> > > >
> > > > GRIB2 is not supported, but I'm thinking that might be very
> outdated??
> > > >
> > > > Thanks for any guidance,
> > > >
> > > > Don
> > > >
> > > > ---
> > > > Don Morton, Owner/Manager
> > > > Boreal Scientific Computing LLC
> > > > Fairbanks, Alaska USA
> > > > http://www.borealscicomp.com/
> > > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> > > >
> > > >
> > >
> > >
> >
> >
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: John Halley Gotway
Time: Fri Nov 04 10:21:42 2016
Don,
I really doubt it's an issue in MET code... and strongly suspect it
has to
with that grib2c library. In fact, I experienced similar behavior in
the
past that I resolved by recompiling that library. So using met-5.1
should
*not* fix the problem, but I've been wrong before!
John
On Fri, Nov 4, 2016 at 10:16 AM, Don Morton via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
>
> Hi John,
>
> I guess I have a five-user limit (including myself), so just removed
two of
> them who don't need access. I'll resend you an invite in a minute.
>
> My container is also CentOS (7), but I complied g2clib from scratch,
so
> I'll try it your way, too. If that fails, maybe I'll backtrack to
Met 5.1,
> though hate to do that.
>
> I'll look at the X stuff later. Seems like something I've run into,
but
> can't remember the resolution now.
>
> Best,
>
> Don
>
> ---
> Don Morton, Owner/Manager
> Boreal Scientific Computing LLC
> Fairbanks, Alaska USA
> http://www.borealscicomp.com/
> http://www.borealscicomp.com/Miscellaneous/MortonBio/
>
> On Fri, Nov 4, 2016 at 3:55 PM, John Halley Gotway via RT <
> met_help at ucar.edu
> > wrote:
>
> > Don,
> >
> > Thanks for the bitbucket invite. Unfortunately, I get this error
> message:
> > The owner of this repository (borealscicomp) is already at
their plan
> > limit and will need to upgrade before adding more users.
> >
> > I'm still very new to container world, but here's the github repo
with
> > version I'm working on:
> > https://github.com/NCAR/container-dtc-met
> >
> > This is the dockerfile, not the pre-built image, but if you have
docker
> > installed, the directions in this readme should enable you to
build the
> > image:
> >
> > https://github.com/NCAR/container-dtc-met/blob/master/
> > met-5.2/README.docker.txt
> >
> > The goal of this container is to make MET easily available for
those who
> > want to go through the online tutorial exercises. So we have them
build
> > the image and then /bin/bash into the environment to go through
the
> > tutorial.
> >
> > However, I still don't have X11 forwarding working... so
displaying GUI's
> > in the container doesn't currently work, such as ncview and gv.
Have you
> > run into similar issues? Any suggestions?
> >
> > As for the GRIB2-C library, I was able to install that as a
package using
> > yum for CentOS:
> > yum -y install g2clib-devel
> >
> > I checked that MET is able to read GRIB2 files in the container
with no
> > segfault. If you're compiling the GRIB2-C library from source,
you could
> > try that package instead to see if you have better luck.
> >
> > John
> >
> > On Thu, Nov 3, 2016 at 6:04 PM, Don Morton via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
> > >
> > > Hi John,
> > >
> > > I just "invited" you (via johnhg at ucar.edu) to share my bitbucket
repo
> > with
> > > the Dockerfile and various files. Should be everything you need
to
> build
> > > and run the image.
> > >
> > > It's very experimental, as my long term goal is to build a "MET
as a
> > > service in AWS" for processing HRRRAK files. To create all the
libs
> and
> > > MET, I fired up a simple container, mounted to one of my host
> > directories,
> > > then did all the compiling/installing in there. The host
directories
> are
> > > now part of the Docker provisioning system, so they get copied
into the
> > > container.
> > >
> > > Not sure I'm really happy with the approach, but I don't want to
go
> > through
> > > a full compilation each time I create a new container.
Tradeoffs... I
> > > think ideally, one might create a MET package that could be
installed
> > with
> > > yum, but I haven't gotten that far.
> > >
> > > I am, indeed, using g2clib 1.6.0 - followed all the instructions
in the
> > MET
> > > v5.2 documentation. But, thanks for the response - I'll poke
around a
> > bit
> > > more. I think I was also having problems with the NAM GRIB2
files...
> > >
> > > Best,
> > >
> > > Don
> > >
> > > ---
> > > Don Morton, Owner/Manager
> > > Boreal Scientific Computing LLC
> > > Fairbanks, Alaska USA
> > > http://www.borealscicomp.com/
> > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> > >
> > > On Thu, Nov 3, 2016 at 10:38 PM, John Halley Gotway via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > > Don,
> > > >
> > > > Yes, GRIB2 inputs should work fine.
> > > >
> > > > Boy you're right, those FAQ's are extremely outdated! That's
the
> > trouble
> > > > with writing something down. You need to keep it updated.
I'll
> make a
> > > > note that we need to update the content there.
> > > >
> > > > My initial guess is that this is related to the version of the
grib2c
> > > > library to which MET is linking. I suspect that recompiling
the
> grib2c
> > > > library and then recompiling MET against that updated library
would
> fix
> > > > it. Try version 1.6.0 of g2clib:
> > > > http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
> > > >
> > > > Please let me know if updating g2clib and recompiling does not
fix
> it.
> > > >
> > > > That's interesting that you're running MET through a docker
> container.
> > > I'm
> > > > working this week on providing exactly that for the MET online
> > tutorial.
> > > > Is that MET docker container available somewhere? Can you
point me
> to
> > > it?
> > > >
> > > > Thanks,
> > > > John Halley Gotway
> > > >
> > > >
> > > > On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT <
> met_help at ucar.edu>
> > > > wrote:
> > > >
> > > > >
> > > > > Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
> > > > > Transaction: Ticket created by don.morton at borealscicomp.com
> > > > > Queue: met_help
> > > > > Subject: Problem running point_stat with HRRR_AK GRIB2
output
> > > > > Owner: Nobody
> > > > > Requestors: don.morton at borealscicomp.com
> > > > > Status: new
> > > > > Ticket <URL: https://rt.rap.ucar.edu/rt/
> > Ticket/Display.html?id=78586
> > > >
> > > > >
> > > > >
> > > > > Hi, (cc'ing a couple of people at ESRL/GSD)
> > > > >
> > > > > I'm running MET 5.2 (provisioned in a Docker Container in
late
> > > September
> > > > -
> > > > > now I see some patches have been issued on 10 October).
But, I'm
> not
> > > > sure
> > > > > that my problem would be related to these.
> > > > >
> > > > > When I run point_stat with the default config file, using a
GRIB2
> > > HRRR_AK
> > > > > (from ESRL/GSD group) I get a seg fault:
> > > > >
> > > > > $ point_stat testdata/1630800001200.wrfprs.gr2
testdata/obs.nc
> > > > > pointstat.conf
> > > > > DEBUG 1: Default Config File:
> > > > > /home/wrfuser/met-
5.2/share/met/config/PointStatConfig_default
> > > > > DEBUG 1: User Config File: pointstat.conf
> > > > > Segmentation fault (core dumped)
> > > > >
> > > > > However, when I use "cnvgrid -g21 ....." to convert the
HRRR-AK to
> > > GRIB1,
> > > > > I'm able to run this successfully and get a reasonable MPR
output
> > file.
> > > > >
> > > > > In looking at documentation, I see that I can't use the "obs
=
> fcst;"
> > > > > statement in the config file if I'm using a non-GRIB1
forecast
> file,
> > > so I
> > > > > know I will need to play around with this. But, in the long
run,
> > > should
> > > > I
> > > > > even expect to be able to run this with GRIB2 input now?
According
> > to
> > > > your
> > > > > FAQ at
> > > > >
> > > > > http://www.dtcenter.org/met/users/support/faqs/index.php
> > > > >
> > > > > GRIB2 is not supported, but I'm thinking that might be very
> > outdated??
> > > > >
> > > > > Thanks for any guidance,
> > > > >
> > > > > Don
> > > > >
> > > > > ---
> > > > > Don Morton, Owner/Manager
> > > > > Boreal Scientific Computing LLC
> > > > > Fairbanks, Alaska USA
> > > > > http://www.borealscicomp.com/
> > > > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: John Halley Gotway
Time: Fri Nov 04 10:47:57 2016
Don,
Thanks for resending the bitbucket link. I was able to access it.
I notice that one big difference is that you're including the source
code
and external libraries directly in the container. In mine, the
Dockerfile
uses "curl" to pull the tarballs from elsewhere. Is there a good
reason to
do it your way?
Also, I just updated the MET Dockerfile to pull a "latest" met-5.2
patch
file rather than having an explicit date listed. The "latest" file is
just
a soft link to the most recent one that I'll update whenever we post
new
patches.
https://github.com/NCAR/container-dtc-met/blob/master/met-
5.2/Dockerfile
John
On Fri, Nov 4, 2016 at 10:21 AM, John Halley Gotway <johnhg at ucar.edu>
wrote:
> Don,
>
> I really doubt it's an issue in MET code... and strongly suspect it
has to
> with that grib2c library. In fact, I experienced similar behavior
in the
> past that I resolved by recompiling that library. So using met-5.1
should
> *not* fix the problem, but I've been wrong before!
>
> John
>
>
> On Fri, Nov 4, 2016 at 10:16 AM, Don Morton via RT
<met_help at ucar.edu>
> wrote:
>
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
>>
>> Hi John,
>>
>> I guess I have a five-user limit (including myself), so just
removed two
>> of
>> them who don't need access. I'll resend you an invite in a minute.
>>
>> My container is also CentOS (7), but I complied g2clib from
scratch, so
>> I'll try it your way, too. If that fails, maybe I'll backtrack to
Met
>> 5.1,
>> though hate to do that.
>>
>> I'll look at the X stuff later. Seems like something I've run
into, but
>> can't remember the resolution now.
>>
>> Best,
>>
>> Don
>>
>> ---
>> Don Morton, Owner/Manager
>> Boreal Scientific Computing LLC
>> Fairbanks, Alaska USA
>> http://www.borealscicomp.com/
>> http://www.borealscicomp.com/Miscellaneous/MortonBio/
>>
>> On Fri, Nov 4, 2016 at 3:55 PM, John Halley Gotway via RT <
>> met_help at ucar.edu
>> > wrote:
>>
>> > Don,
>> >
>> > Thanks for the bitbucket invite. Unfortunately, I get this error
>> message:
>> > The owner of this repository (borealscicomp) is already at
their plan
>> > limit and will need to upgrade before adding more users.
>> >
>> > I'm still very new to container world, but here's the github repo
with
>> > version I'm working on:
>> > https://github.com/NCAR/container-dtc-met
>> >
>> > This is the dockerfile, not the pre-built image, but if you have
docker
>> > installed, the directions in this readme should enable you to
build the
>> > image:
>> >
>> > https://github.com/NCAR/container-dtc-met/blob/master/
>> > met-5.2/README.docker.txt
>> >
>> > The goal of this container is to make MET easily available for
those who
>> > want to go through the online tutorial exercises. So we have
them build
>> > the image and then /bin/bash into the environment to go through
the
>> > tutorial.
>> >
>> > However, I still don't have X11 forwarding working... so
displaying
>> GUI's
>> > in the container doesn't currently work, such as ncview and gv.
Have
>> you
>> > run into similar issues? Any suggestions?
>> >
>> > As for the GRIB2-C library, I was able to install that as a
package
>> using
>> > yum for CentOS:
>> > yum -y install g2clib-devel
>> >
>> > I checked that MET is able to read GRIB2 files in the container
with no
>> > segfault. If you're compiling the GRIB2-C library from source,
you
>> could
>> > try that package instead to see if you have better luck.
>> >
>> > John
>> >
>> > On Thu, Nov 3, 2016 at 6:04 PM, Don Morton via RT
<met_help at ucar.edu>
>> > wrote:
>> >
>> > >
>> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
>> > >
>> > > Hi John,
>> > >
>> > > I just "invited" you (via johnhg at ucar.edu) to share my
bitbucket repo
>> > with
>> > > the Dockerfile and various files. Should be everything you
need to
>> build
>> > > and run the image.
>> > >
>> > > It's very experimental, as my long term goal is to build a "MET
as a
>> > > service in AWS" for processing HRRRAK files. To create all the
libs
>> and
>> > > MET, I fired up a simple container, mounted to one of my host
>> > directories,
>> > > then did all the compiling/installing in there. The host
directories
>> are
>> > > now part of the Docker provisioning system, so they get copied
into
>> the
>> > > container.
>> > >
>> > > Not sure I'm really happy with the approach, but I don't want
to go
>> > through
>> > > a full compilation each time I create a new container.
Tradeoffs...
>> I
>> > > think ideally, one might create a MET package that could be
installed
>> > with
>> > > yum, but I haven't gotten that far.
>> > >
>> > > I am, indeed, using g2clib 1.6.0 - followed all the
instructions in
>> the
>> > MET
>> > > v5.2 documentation. But, thanks for the response - I'll poke
around a
>> > bit
>> > > more. I think I was also having problems with the NAM GRIB2
files...
>> > >
>> > > Best,
>> > >
>> > > Don
>> > >
>> > > ---
>> > > Don Morton, Owner/Manager
>> > > Boreal Scientific Computing LLC
>> > > Fairbanks, Alaska USA
>> > > http://www.borealscicomp.com/
>> > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
>> > >
>> > > On Thu, Nov 3, 2016 at 10:38 PM, John Halley Gotway via RT <
>> > > met_help at ucar.edu> wrote:
>> > >
>> > > > Don,
>> > > >
>> > > > Yes, GRIB2 inputs should work fine.
>> > > >
>> > > > Boy you're right, those FAQ's are extremely outdated! That's
the
>> > trouble
>> > > > with writing something down. You need to keep it updated.
I'll
>> make a
>> > > > note that we need to update the content there.
>> > > >
>> > > > My initial guess is that this is related to the version of
the
>> grib2c
>> > > > library to which MET is linking. I suspect that recompiling
the
>> grib2c
>> > > > library and then recompiling MET against that updated library
would
>> fix
>> > > > it. Try version 1.6.0 of g2clib:
>> > > > http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
>> > > >
>> > > > Please let me know if updating g2clib and recompiling does
not fix
>> it.
>> > > >
>> > > > That's interesting that you're running MET through a docker
>> container.
>> > > I'm
>> > > > working this week on providing exactly that for the MET
online
>> > tutorial.
>> > > > Is that MET docker container available somewhere? Can you
point me
>> to
>> > > it?
>> > > >
>> > > > Thanks,
>> > > > John Halley Gotway
>> > > >
>> > > >
>> > > > On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT <
>> met_help at ucar.edu>
>> > > > wrote:
>> > > >
>> > > > >
>> > > > > Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
>> > > > > Transaction: Ticket created by don.morton at borealscicomp.com
>> > > > > Queue: met_help
>> > > > > Subject: Problem running point_stat with HRRR_AK GRIB2
output
>> > > > > Owner: Nobody
>> > > > > Requestors: don.morton at borealscicomp.com
>> > > > > Status: new
>> > > > > Ticket <URL: https://rt.rap.ucar.edu/rt/
>> > Ticket/Display.html?id=78586
>> > > >
>> > > > >
>> > > > >
>> > > > > Hi, (cc'ing a couple of people at ESRL/GSD)
>> > > > >
>> > > > > I'm running MET 5.2 (provisioned in a Docker Container in
late
>> > > September
>> > > > -
>> > > > > now I see some patches have been issued on 10 October).
But, I'm
>> not
>> > > > sure
>> > > > > that my problem would be related to these.
>> > > > >
>> > > > > When I run point_stat with the default config file, using a
GRIB2
>> > > HRRR_AK
>> > > > > (from ESRL/GSD group) I get a seg fault:
>> > > > >
>> > > > > $ point_stat testdata/1630800001200.wrfprs.gr2
testdata/obs.nc
>> > > > > pointstat.conf
>> > > > > DEBUG 1: Default Config File:
>> > > > > /home/wrfuser/met-
5.2/share/met/config/PointStatConfig_default
>> > > > > DEBUG 1: User Config File: pointstat.conf
>> > > > > Segmentation fault (core dumped)
>> > > > >
>> > > > > However, when I use "cnvgrid -g21 ....." to convert the
HRRR-AK to
>> > > GRIB1,
>> > > > > I'm able to run this successfully and get a reasonable MPR
output
>> > file.
>> > > > >
>> > > > > In looking at documentation, I see that I can't use the
"obs =
>> fcst;"
>> > > > > statement in the config file if I'm using a non-GRIB1
forecast
>> file,
>> > > so I
>> > > > > know I will need to play around with this. But, in the
long run,
>> > > should
>> > > > I
>> > > > > even expect to be able to run this with GRIB2 input now?
>> According
>> > to
>> > > > your
>> > > > > FAQ at
>> > > > >
>> > > > > http://www.dtcenter.org/met/users/support/faqs/index.php
>> > > > >
>> > > > > GRIB2 is not supported, but I'm thinking that might be very
>> > outdated??
>> > > > >
>> > > > > Thanks for any guidance,
>> > > > >
>> > > > > Don
>> > > > >
>> > > > > ---
>> > > > > Don Morton, Owner/Manager
>> > > > > Boreal Scientific Computing LLC
>> > > > > Fairbanks, Alaska USA
>> > > > > http://www.borealscicomp.com/
>> > > > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
>> > > > >
>> > > > >
>> > > >
>> > > >
>> > >
>> > >
>> >
>> >
>>
>>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: Don Morton
Time: Fri Nov 04 10:58:56 2016
Hi John,
My logic (the logic of an old, scatterbrained computer scientist) was
that
I wanted to be able to version control the full package and not
necessarily
rely on packages that may or may not be available on the web.
I kept the source and such just because I'm still experimenting, and
didn't
want to trash it all just yet. Eventually I should be able to clean
it out
a lot. Frankly, there really shouldn't even be a need for compilers
once
everything is in place.
Given your recent comments about use of the CentOS native g2clib, I'm
thinking of doing a new container where I actually just compile it all
as
part of the provisioning. Yeh, it'll take longer, but right now I'm
not
comfortable with the idea that my executables require external
processes
for their creation, and if I want to change something, I have to go
back
and recompile the directories that I import into Docker.
But, I'm still learning... a lot!! I'm also working with vagrant to
create
VMs for both VirtualBox and AWS - that's more of a WRF deployment...
Best,
Don
---
Don Morton, Owner/Manager
Boreal Scientific Computing LLC
Fairbanks, Alaska USA
http://www.borealscicomp.com/
http://www.borealscicomp.com/Miscellaneous/MortonBio/
On Fri, Nov 4, 2016 at 4:47 PM, John Halley Gotway via RT
<met_help at ucar.edu
> wrote:
> Don,
>
> Thanks for resending the bitbucket link. I was able to access it.
>
> I notice that one big difference is that you're including the source
code
> and external libraries directly in the container. In mine, the
Dockerfile
> uses "curl" to pull the tarballs from elsewhere. Is there a good
reason to
> do it your way?
>
> Also, I just updated the MET Dockerfile to pull a "latest" met-5.2
patch
> file rather than having an explicit date listed. The "latest" file
is just
> a soft link to the most recent one that I'll update whenever we post
new
> patches.
>
> https://github.com/NCAR/container-dtc-met/blob/master/met-
5.2/Dockerfile
>
> John
>
>
> On Fri, Nov 4, 2016 at 10:21 AM, John Halley Gotway
<johnhg at ucar.edu>
> wrote:
>
> > Don,
> >
> > I really doubt it's an issue in MET code... and strongly suspect
it has
> to
> > with that grib2c library. In fact, I experienced similar behavior
in the
> > past that I resolved by recompiling that library. So using met-
5.1
> should
> > *not* fix the problem, but I've been wrong before!
> >
> > John
> >
> >
> > On Fri, Nov 4, 2016 at 10:16 AM, Don Morton via RT
<met_help at ucar.edu>
> > wrote:
> >
> >>
> >> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
> >>
> >> Hi John,
> >>
> >> I guess I have a five-user limit (including myself), so just
removed two
> >> of
> >> them who don't need access. I'll resend you an invite in a
minute.
> >>
> >> My container is also CentOS (7), but I complied g2clib from
scratch, so
> >> I'll try it your way, too. If that fails, maybe I'll backtrack
to Met
> >> 5.1,
> >> though hate to do that.
> >>
> >> I'll look at the X stuff later. Seems like something I've run
into, but
> >> can't remember the resolution now.
> >>
> >> Best,
> >>
> >> Don
> >>
> >> ---
> >> Don Morton, Owner/Manager
> >> Boreal Scientific Computing LLC
> >> Fairbanks, Alaska USA
> >> http://www.borealscicomp.com/
> >> http://www.borealscicomp.com/Miscellaneous/MortonBio/
> >>
> >> On Fri, Nov 4, 2016 at 3:55 PM, John Halley Gotway via RT <
> >> met_help at ucar.edu
> >> > wrote:
> >>
> >> > Don,
> >> >
> >> > Thanks for the bitbucket invite. Unfortunately, I get this
error
> >> message:
> >> > The owner of this repository (borealscicomp) is already at
their
> plan
> >> > limit and will need to upgrade before adding more users.
> >> >
> >> > I'm still very new to container world, but here's the github
repo with
> >> > version I'm working on:
> >> > https://github.com/NCAR/container-dtc-met
> >> >
> >> > This is the dockerfile, not the pre-built image, but if you
have
> docker
> >> > installed, the directions in this readme should enable you to
build
> the
> >> > image:
> >> >
> >> > https://github.com/NCAR/container-dtc-met/blob/master/
> >> > met-5.2/README.docker.txt
> >> >
> >> > The goal of this container is to make MET easily available for
those
> who
> >> > want to go through the online tutorial exercises. So we have
them
> build
> >> > the image and then /bin/bash into the environment to go through
the
> >> > tutorial.
> >> >
> >> > However, I still don't have X11 forwarding working... so
displaying
> >> GUI's
> >> > in the container doesn't currently work, such as ncview and gv.
Have
> >> you
> >> > run into similar issues? Any suggestions?
> >> >
> >> > As for the GRIB2-C library, I was able to install that as a
package
> >> using
> >> > yum for CentOS:
> >> > yum -y install g2clib-devel
> >> >
> >> > I checked that MET is able to read GRIB2 files in the container
with
> no
> >> > segfault. If you're compiling the GRIB2-C library from source,
you
> >> could
> >> > try that package instead to see if you have better luck.
> >> >
> >> > John
> >> >
> >> > On Thu, Nov 3, 2016 at 6:04 PM, Don Morton via RT
<met_help at ucar.edu>
> >> > wrote:
> >> >
> >> > >
> >> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586
>
> >> > >
> >> > > Hi John,
> >> > >
> >> > > I just "invited" you (via johnhg at ucar.edu) to share my
bitbucket
> repo
> >> > with
> >> > > the Dockerfile and various files. Should be everything you
need to
> >> build
> >> > > and run the image.
> >> > >
> >> > > It's very experimental, as my long term goal is to build a
"MET as a
> >> > > service in AWS" for processing HRRRAK files. To create all
the libs
> >> and
> >> > > MET, I fired up a simple container, mounted to one of my host
> >> > directories,
> >> > > then did all the compiling/installing in there. The host
> directories
> >> are
> >> > > now part of the Docker provisioning system, so they get
copied into
> >> the
> >> > > container.
> >> > >
> >> > > Not sure I'm really happy with the approach, but I don't want
to go
> >> > through
> >> > > a full compilation each time I create a new container.
Tradeoffs...
> >> I
> >> > > think ideally, one might create a MET package that could be
> installed
> >> > with
> >> > > yum, but I haven't gotten that far.
> >> > >
> >> > > I am, indeed, using g2clib 1.6.0 - followed all the
instructions in
> >> the
> >> > MET
> >> > > v5.2 documentation. But, thanks for the response - I'll poke
> around a
> >> > bit
> >> > > more. I think I was also having problems with the NAM GRIB2
> files...
> >> > >
> >> > > Best,
> >> > >
> >> > > Don
> >> > >
> >> > > ---
> >> > > Don Morton, Owner/Manager
> >> > > Boreal Scientific Computing LLC
> >> > > Fairbanks, Alaska USA
> >> > > http://www.borealscicomp.com/
> >> > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> >> > >
> >> > > On Thu, Nov 3, 2016 at 10:38 PM, John Halley Gotway via RT <
> >> > > met_help at ucar.edu> wrote:
> >> > >
> >> > > > Don,
> >> > > >
> >> > > > Yes, GRIB2 inputs should work fine.
> >> > > >
> >> > > > Boy you're right, those FAQ's are extremely outdated!
That's the
> >> > trouble
> >> > > > with writing something down. You need to keep it updated.
I'll
> >> make a
> >> > > > note that we need to update the content there.
> >> > > >
> >> > > > My initial guess is that this is related to the version of
the
> >> grib2c
> >> > > > library to which MET is linking. I suspect that
recompiling the
> >> grib2c
> >> > > > library and then recompiling MET against that updated
library
> would
> >> fix
> >> > > > it. Try version 1.6.0 of g2clib:
> >> > > > http://www.nco.ncep.noaa.gov/pmb/codes/GRIB2/
> >> > > >
> >> > > > Please let me know if updating g2clib and recompiling does
not fix
> >> it.
> >> > > >
> >> > > > That's interesting that you're running MET through a docker
> >> container.
> >> > > I'm
> >> > > > working this week on providing exactly that for the MET
online
> >> > tutorial.
> >> > > > Is that MET docker container available somewhere? Can you
point
> me
> >> to
> >> > > it?
> >> > > >
> >> > > > Thanks,
> >> > > > John Halley Gotway
> >> > > >
> >> > > >
> >> > > > On Thu, Nov 3, 2016 at 12:58 PM, Don Morton via RT <
> >> met_help at ucar.edu>
> >> > > > wrote:
> >> > > >
> >> > > > >
> >> > > > > Thu Nov 03 12:58:22 2016: Request 78586 was acted upon.
> >> > > > > Transaction: Ticket created by
don.morton at borealscicomp.com
> >> > > > > Queue: met_help
> >> > > > > Subject: Problem running point_stat with HRRR_AK
GRIB2
> output
> >> > > > > Owner: Nobody
> >> > > > > Requestors: don.morton at borealscicomp.com
> >> > > > > Status: new
> >> > > > > Ticket <URL: https://rt.rap.ucar.edu/rt/
> >> > Ticket/Display.html?id=78586
> >> > > >
> >> > > > >
> >> > > > >
> >> > > > > Hi, (cc'ing a couple of people at ESRL/GSD)
> >> > > > >
> >> > > > > I'm running MET 5.2 (provisioned in a Docker Container in
late
> >> > > September
> >> > > > -
> >> > > > > now I see some patches have been issued on 10 October).
But,
> I'm
> >> not
> >> > > > sure
> >> > > > > that my problem would be related to these.
> >> > > > >
> >> > > > > When I run point_stat with the default config file, using
a
> GRIB2
> >> > > HRRR_AK
> >> > > > > (from ESRL/GSD group) I get a seg fault:
> >> > > > >
> >> > > > > $ point_stat testdata/1630800001200.wrfprs.gr2
testdata/obs.nc
> >> > > > > pointstat.conf
> >> > > > > DEBUG 1: Default Config File:
> >> > > > > /home/wrfuser/met-
5.2/share/met/config/PointStatConfig_default
> >> > > > > DEBUG 1: User Config File: pointstat.conf
> >> > > > > Segmentation fault (core dumped)
> >> > > > >
> >> > > > > However, when I use "cnvgrid -g21 ....." to convert the
HRRR-AK
> to
> >> > > GRIB1,
> >> > > > > I'm able to run this successfully and get a reasonable
MPR
> output
> >> > file.
> >> > > > >
> >> > > > > In looking at documentation, I see that I can't use the
"obs =
> >> fcst;"
> >> > > > > statement in the config file if I'm using a non-GRIB1
forecast
> >> file,
> >> > > so I
> >> > > > > know I will need to play around with this. But, in the
long
> run,
> >> > > should
> >> > > > I
> >> > > > > even expect to be able to run this with GRIB2 input now?
> >> According
> >> > to
> >> > > > your
> >> > > > > FAQ at
> >> > > > >
> >> > > > > http://www.dtcenter.org/met/users/support/faqs/index.php
> >> > > > >
> >> > > > > GRIB2 is not supported, but I'm thinking that might be
very
> >> > outdated??
> >> > > > >
> >> > > > > Thanks for any guidance,
> >> > > > >
> >> > > > > Don
> >> > > > >
> >> > > > > ---
> >> > > > > Don Morton, Owner/Manager
> >> > > > > Boreal Scientific Computing LLC
> >> > > > > Fairbanks, Alaska USA
> >> > > > > http://www.borealscicomp.com/
> >> > > > > http://www.borealscicomp.com/Miscellaneous/MortonBio/
> >> > > > >
> >> > > > >
> >> > > >
> >> > > >
> >> > >
> >> > >
> >> >
> >> >
> >>
> >>
> >
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: Don Morton
Time: Fri Nov 04 13:04:24 2016
On Fri, Nov 4, 2016 at 3:55 PM, John Halley Gotway via RT
<met_help at ucar.edu
> wrote:
>
>
> As for the GRIB2-C library, I was able to install that as a package
using
> yum for CentOS:
> yum -y install g2clib-devel
>
>
Hi, this fixed the problem for me. Interestingly, the CentOS7 version
of
g2clib is 1.4.
For my European projects, I use ECMWF's grib_api for some of the same
things, and they seemed to have several bad versions (only under some
unique cases) where people just regularly ended up using the older
versions
until the problems were finally fixed.
Thanks for the hint. I was trying to use strace to find the source of
the
problems, and was sinking very deep into a black hole. You save me
before
I passed the event horizon!
Best,
Don
>
>
------------------------------------------------
Subject: Problem running point_stat with HRRR_AK GRIB2 output
From: John Halley Gotway
Time: Fri Nov 04 13:40:36 2016
Thanks for letting me know that did the trick.
I'll go ahead and resolve the issue.
Thanks,
John
On Fri, Nov 4, 2016 at 1:04 PM, Don Morton via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=78586 >
>
> On Fri, Nov 4, 2016 at 3:55 PM, John Halley Gotway via RT <
> met_help at ucar.edu
> > wrote:
>
> >
> >
> > As for the GRIB2-C library, I was able to install that as a
package using
> > yum for CentOS:
> > yum -y install g2clib-devel
> >
> >
>
> Hi, this fixed the problem for me. Interestingly, the CentOS7
version of
> g2clib is 1.4.
>
> For my European projects, I use ECMWF's grib_api for some of the
same
> things, and they seemed to have several bad versions (only under
some
> unique cases) where people just regularly ended up using the older
versions
> until the problems were finally fixed.
>
> Thanks for the hint. I was trying to use strace to find the source
of the
> problems, and was sinking very deep into a black hole. You save me
before
> I passed the event horizon!
>
> Best,
>
> Don
>
> >
> >
>
>
------------------------------------------------
More information about the Met_help
mailing list