RAMONA mosaic UDP fails for South Africa

[Patrick1G]

Previously this job failed due to the island polygon of Lesotho, but that has been fixed:
Open-EO/openeo-python-driver#433

Now attempted to rerun and job fails after processing for a good amount of time:
j-2603051049014aee8343c76b7f3c8e46

A lot of error messages in the log. Would be good to know what happened here..
@jdries

https://algorithm-catalogue.apex.esa.int/apps/RAMONA-herbaceous_rangeland_biomass-country-mosaick

[jdries]

This is most likely memory (job option) related. I was doing a few runs to give you the solution, but each one is taking a rather long time. So solution should be underway, do let me know if we should raise priority.

[jdries]

Ok, so now it came through, this is the largest ramona output I've seen so far, with a size of 230597x265868, 16,9GB.

Options used:
'driver-memory': '19G', 'executor-memory': '8G', 'python-memory': '50m', 'executor-memoryOverhead': '3g', 'driver-memoryOverhead': '5G'

https://openeo.dataspace.copernicus.eu/openeo/1.2/jobs/j-260312094024492fac7be0c082d63570/results/MzJjYzdkZGItZjdlMS00YjFjLTk3OTYtZjlmZTM5Y2I4ZmVi/4e2e6730c09d8168589bce4d9248b9e8?expires=1773921318

Shall I update the UDP, or is it ok to only use these extended options for the very large countries?

[Patrick1G]

good question @jdries

i guess enabling these options would lead to larger resource consumption/allocation anytime this UDP is run?

if that is not the case, then lets certainly enable those options by default..

Ideally the UDP would be smart and apply those options once more than 100,00KM2...

[Patrick1G]

@jdries @jamesemwheeler

just a few observations on the above dataset:

• no data value is set to -32768 while the individual tiles have 0 as no data value
• gdalinfo check looks ok, none of the previously observed issues
• when visualising in QGIS there are a number of issues related to read block errors (3 examples below see below)
• also in QGIS when zooming in close, e.g. 1:25000 the rendering fails.. 1:50000 still works.. some issue with the pyramids?

Would be good to check this one also with the APEx GE and check COG compliance @jamesemwheeler

ramona_hrb_kenya_03-26: IReadBlock failed at X offset 97, Y offset 109: TIFFReadEncodedTile() failed.
ramona_hrb_kenya_03-26: IReadBlock failed at X offset 97, Y offset 109: TIFFReadEncodedTile() failed.
ramona_hrb_kenya_03-26: ramona_hrb_kenya_03-26.tif, band 1: IReadBlock failed at X offset 166, Y offset 210: TIFFReadEncodedTile() failed.

% gdalinfo ramona_hrb_kenya_03-26.tif
Driver: GTiff/GeoTIFF
Files: ramona_hrb_kenya_03-26.tif
Size is 230597, 265868
Coordinate System is:
GEOGCRS["WGS 84",
DATUM["World Geodetic System 1984",
ELLIPSOID["WGS 84",6378137,298.257223563,
LENGTHUNIT["metre",1,
ID["EPSG",9001]]]],
PRIMEM["Greenwich",0,
ANGLEUNIT["degree",0.0174532925199433,
ID["EPSG",9122]]],
CS[ellipsoidal,2],
AXIS["latitude",north,
ORDER[1],
ANGLEUNIT["degree",0.0174532925199433,
ID["EPSG",9122]]],
AXIS["longitude",east,
ORDER[2],
ANGLEUNIT["degree",0.0174532925199433,
ID["EPSG",9122]]]]
Data axis to CRS axis mapping: 2,1
Origin = (16.450651314486429,-22.123728816674518)
Pixel Size = (0.000093558980351,-0.000093558973882)
Metadata:
institution=Copernicus Data Space Ecosystem openEO API - 0.71.0a19.dev20260310+3480
PROCESSING_SOFTWARE=0.71.0a19
AREA_OR_POINT=Area
Image Structure Metadata:
COMPRESSION=DEFLATE
INTERLEAVE=BAND
Corner Coordinates:
Upper Left ( 16.4506513, -22.1237288) ( 16d27' 2.34"E, 22d 7'25.42"S)
Lower Left ( 16.4506513, -46.9980661) ( 16d27' 2.34"E, 46d59'53.04"S)
Upper Right ( 38.0250715, -22.1237288) ( 38d 1'30.26"E, 22d 7'25.42"S)
Lower Right ( 38.0250715, -46.9980661) ( 38d 1'30.26"E, 46d59'53.04"S)
Center ( 27.2378614, -34.5608975) ( 27d14'16.30"E, 34d33'39.23"S)
Band 1 Block=512x512 Type=Int16, ColorInterp=Gray
Description = biomass
NoData Value=-32768
Overviews: 115299x132934, 57650x66467, 28825x33234, 14413x16617, 7207x8309, 3604x4155, 1802x2078, 901x1039, 451x520, 226x260
Offset: 0, Scale:0.01
Metadata:
UNIT=g DW per m2 per month