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
AFAICT, it's currently required to specify exactly one authentication type in fs.gs.auth.type. This is fine for clusters, but annoying for non-expert users who go back and forth between local prototyping on a laptop and execution on a cluster. For those non-expert users, they have to figure out where spark-defaults.conf is stored and edit that file. To edit that file they have to further know that the configuration variable must be preceded by spark.hadoop..
In contrast, the google-auth Python library provides google.auth.default() which allows users and libraries to request credentials in an environment independent way. It just "does the right thing".
Concretely the request is for:
A "DEFAULT" option to fs.gs.auth.type which uses the exact same precedence as google.auth.default(), and
That the default setting for fs.gs.auth.type be "DEFAULT"
The text was updated successfully, but these errors were encountered:
AFAICT, it's currently required to specify exactly one authentication type in
fs.gs.auth.type
. This is fine for clusters, but annoying for non-expert users who go back and forth between local prototyping on a laptop and execution on a cluster. For those non-expert users, they have to figure out where spark-defaults.conf is stored and edit that file. To edit that file they have to further know that the configuration variable must be preceded byspark.hadoop.
.In contrast, the
google-auth
Python library providesgoogle.auth.default()
which allows users and libraries to request credentials in an environment independent way. It just "does the right thing".Concretely the request is for:
"DEFAULT"
option tofs.gs.auth.type
which uses the exact same precedence asgoogle.auth.default()
, andfs.gs.auth.type
be"DEFAULT"
The text was updated successfully, but these errors were encountered: