You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Over the past few months we've been encountering intermittent GDAL access issue semi-regularly. e.g.:
CPLE_OpenFailedError: '/vsis3/dea-public-data/baseline/ga_s2bm_ard_3/52/LBK/2020/08/15/20200815T032931/ga_s2bm_nbart_3-2-1_52LBK_2020-08-15_final_band03.tif' not recognized as a supported file format.
This is a real pain, particularly in automated testing where a random fail can cause us to need to re-run our entire slow test suite.
datacube.load has a handy skip_broken_datasets param that can be used to workaround this issue. However, we don't really want to set this in every notebook/script as it adds complexity and potentially makes workflows non-reproducible.
Thougts on adding support for a global environmental variable (e.g. ODC_SKIP_BROKEN_DATASETS etc) that could be set to force datacube to skip broken datasets, even if this wasn't set in Python code itself? This would allow us to set this in our tests, allowing the tests to be more robust to these issues without impacting user code.
The text was updated successfully, but these errors were encountered:
Probably best handled in the 1.9 branch after #1505 is merged.
Add skip_broken_datasets as a config option, defaulting to False. Will automatically be over-rideable per environment with e.g. ODC_PROD_SKIP_BROKEN_DATASETS and/or ODC_DEV_SKIP_BROKEN_DATASETS.
Over the past few months we've been encountering intermittent GDAL access issue semi-regularly. e.g.:
This is a real pain, particularly in automated testing where a random fail can cause us to need to re-run our entire slow test suite.
datacube.load
has a handyskip_broken_datasets
param that can be used to workaround this issue. However, we don't really want to set this in every notebook/script as it adds complexity and potentially makes workflows non-reproducible.Thougts on adding support for a global environmental variable (e.g.
ODC_SKIP_BROKEN_DATASETS
etc) that could be set to force datacube to skip broken datasets, even if this wasn't set in Python code itself? This would allow us to set this in our tests, allowing the tests to be more robust to these issues without impacting user code.The text was updated successfully, but these errors were encountered: