RE: [las_users]

Hi.

We see this behavior when we ask it to do a Hovmoller plot or any plot that spans multiple years. Today our servers aren't seeing as much use, and the Hovmoller for ODYSSEA barely squeaked to finish right before the 300 second mark -- most of the time, it can't complete 1 year of data by 300 seconds.

Here's an example using OSTIA; it, too, has the same behavior but spans a longer time and thus is sure to not complete in 300 seconds.

goes to

The link to OSTIA is here:
http://thredds.jpl.nasa.gov/thredds/dodsC/sea_surface_temperature/ALL_UKMO-L4HRfnd-GLOB-OSTIA_v01-fv02.nc.html

Here is an example of trying to do a time series:

Gives:

Meanwhile I can watch the logs scroll along as the backend continues to open up the rest of the granules in the request.

The exact time span that completes in under 300 seconds varies depending on how much traffic the site is receiving; and of course, all the job requests where the front end fails after 300 seconds are still there and contributing to the load in the background, making the next request even more tight on the time-range that will complete.

Thank you for looking into this.

Kacie

From: Roland Schweitzer - NOAA Affiliate [roland.schweitzer@xxxxxxxx]
Sent: Friday, April 05, 2013 12:46 PM
To: Shelton, Kacie E (398C)
Cc: las_users@xxxxxxxx; Sumagaysay, Rosanna M (3980-Affiliate)
Subject: Re: [las_users]

Hi,

Now I'm really confused.

This data set works like a champ though there seems to be some missing data in the middle. I'm making plots of all the different variables and the first, last and some random time steps in between and they all work and return in a matter of a couple of seconds.

And this is the full data set that fails after 300 seconds on your server?

Roland

On Fri, Apr 5, 2013 at 12:58 PM, Shelton, Kacie E (398C) <kacie.e.shelton@xxxxxxxxxxxx> wrote:

Hi.

Try this link:

http://thredds.jpl.nasa.gov/thredds/dodsC/GHRSST_EUR-L4HRfnd-GLOB-ODYSSEA_TIMEAGG/aggregate__ghrsst_EUR-L4HRfnd-GLOB-ODYSSEA.ncml.html

I accidentally sent you an older link in our testbed. Sorry!

We will look at our tomcat server files and see if there's a parameter there we've missed.

Thank you!

Kacie

From: Roland Schweitzer - NOAA Affiliate [roland.schweitzer@xxxxxxxx]

Sent: Friday, April 05, 2013 7:04 AM
To: Shelton, Kacie E (398C)
Cc: las_users@xxxxxxxx; Sumagaysay, Rosanna M (3980-Affiliate)
Subject: Re: [las_users]

I configured the GHRSST data into my LAS, but I cannot at the moment reach the data at this URL:

http://seastore.jpl.nasa.gov/thredds/dodsC/GHRSST_EUR-L4HRfnd-GLOB-ODYSSEA_TIMEAGG/aggregate__ghrsst_EUR-L4HRfnd-GLOB-ODYSSEA.ncml.html

Is this server behind a firewall, or just down at the moment?

In LAS Ferret runs as the same user as the tomcat server. Does that user have an operating system limit on the number of CPU seconds it can spend on a single process?

If these data are public and I can get access to a running server, I will try it in LAS from here.

Roland

On Mon, Apr 1, 2013 at 1:23 PM, Shelton, Kacie E (388C) <kacie.e.shelton@xxxxxxxxxxxx> wrote:

Hi.

Thanks Ansley, we fixed our display issue by manually sourcing the ferret_paths_template.sh file in /usr/local/ferret/bin -- we got the plot below.

We are still having the disconnect/time-out issue, happening at 300 seconds (5min) between the front-end & back-end.

Any ideas there, Roland?

Thanks for all your help; this issue is a persistent thorn for us.

Kacie Shelton

From: Roland Schweitzer - NOAA Affiliate [roland.schweitzer@xxxxxxxx]
Sent: Tuesday, March 26, 2013 7:19 AM

To: Shelton, Kacie E (388C)
Cc: las_users@xxxxxxxx; Sumagaysay, Rosanna M (3880-Affiliate)
Subject: Re: [las_users]

I was able to make a Hovmoller plot on my LAS on my laptop for this data set over the full time range. Perhaps this is not a good test.

Can you please send me the data set configuration you are using for a data set that is not working so I can test with that?

Thanks,

Roland

On Tue, Mar 26, 2013 at 9:03 AM, Roland Schweitzer - NOAA Affiliate <roland.schweitzer@xxxxxxxx> wrote:

I think you have an issue with the setup of your Ferret environment variables for your attempts to make a plot from the command line. In particular the value of FER_PALETTE= which should point to a directory that contains the default palette.

I will try setting up my LAS using one of your large data sets and see if I can figure out what's going on.

I'm trying this one:

http://thredds.jpl.nasa.gov/thredds/dodsC/MODIS_AQUA_L3_SST_THERMAL_DAILY_4KM_DAYTIME_TIMEAGG/aggregate__MODIS_AQUA_L3_SST_THERMAL_DAILY_4KM_DAYTIME.ncml.html

Is that one of the ones that fails on a long Hovmoller plot?

Roland

On Mon, Mar 25, 2013 at 12:29 PM, Shelton, Kacie E (388C) <kacie.e.shelton@xxxxxxxxxxxx> wrote:

Hello, Roland;

We are still having an issue with our front-end timing out and dying at 5min, while the back-end continues to chug along trying to complete the request. Even if the back-end process completes successfully, the results are not passed to the front-end.

We tried running ferret directly as you suggested, to see if we could get a plot, but we are getting an error. (Results below.)

Do you (or anyone on the list) have any ideas?

Thank you,
Kacie Shelton

-----

Our LAS server is public:   http://thredds.jpl.nasa.gov/las/getUI.do

When Rosanna and I run the code in ferret, we both get:

PPL+ error: Unable to read "default.spk" to set shade and fill colors.

PPL+ error: Unable to read "default.spk" to set shade and fill colors.

At line 171 of file symbel.f

Fortran runtime error: No such file or directory

Our machines have g77 (gfortran) on them. Is that part of the error, or a separate issue?

The dataset http://thredds.jpl.nasa.gov/thredds/dodsC/GHRSST_EUR-L4HRfnd-GLOB-ODYSSEA_TIMEAGG/aggregate__ghrsst_EUR-L4HRfnd-GLOB-ODYSSEA.ncml is big enough that a Hovmoller longitude plot doesn’t complete when set to the full time range in LAS – it will complete about 11months, but longer than that and we see the 300 second time out, while as you say, OPeNDAP chugs along its merry way even though Ferret has timed out. (ODYSSEA is not as big as other datasets, so at least it doesn’t tie up OPeNDAP as long.)

Would you please give it a try on your end, in case the fortran issue is a separate problem?

My attempt is below; please ignore the xterm issue, as that’s my windows machine. Rosanna gets the Fortran error, but not the xterm display issue.

Thank you,

Kacie Shelton

[kshelton@seastore ~]$ ferret

        NOAA/PMEL TMAP

        FERRET v6.72

        Linux 2.6.18-274.el5 64-bit - 09/13/11

        15-Mar-13 15:46

yes? use "http://thredds.jpl.nasa.gov/thredds/dodsC/GHRSST_EUR-L4HRfnd-GLOB-ODYSSEA_TIMEAGG/aggregate__ghrsst_EUR-L4HRfnd-GLOB-ODYSSEA.ncml"

yes? show data/all

     currently SET data sets:

    1> http://thredds.jpl.nasa.gov/thredds/dodsC/GHRSST_EUR-L4HRfnd-GLOB-ODYSSEA_TIMEAGG/aggregate__ghrsst_EUR-L4HRfnd-GLOB-ODYSSEA.ncml (default)

name     title                             I         J         K         L

ANALYSED_SST

          analysed sea surface temperatur 1:3600    1:1600    ...       1:609

ANALYSIS_ERROR

          estimated error percentage       1:3600    1:1600    ...       1:609

MASK     sea/land/lake/ice field composi 1:3600    1:1600    ...       1:609

SEA_ICE_FRACTION

          sea ice area fraction            1:3600    1:1600    ...       1:609

yes? contour/l=1 ANALYSED_SST

PPL+ error: Unable to read "default.spk" to set shade and fill colors.

Can't connect to X Server :0

Your DISPLAY environment variable must be set to

point to a working X server.

gopenws()   26 Specified workstation cannot be opened

gactivatews()    6 GKS not in proper state: GKS shall be either in the state WSOP or in the state WSAC

gsetdeferst()    7 GKS not in proper state: GKS shall be in one of the states WSOP, WSAC or SGOP

gsetlinerep()    7 GKS not in proper state: GKS shall be in one of the states WSOP, WSAC or SGOP

ginqmaxwssttables()   22 Specified workstation type is invalid

PPL+ error: Unable to read "default.spk" to set shade and fill colors.

At line 171 of file symbel.f

Fortran runtime error: No such file or directory