in some cases user might not want to keep on retry-ing upload forever.
the logic makes sense for "transient" errors (e.g. network/service failure) but is potentially problematic (generating a lot of noise at minimum) for arbitrary raises.
we potentially need to investigate the class of exceptions coming from s3 library.
in some cases user might not want to keep on retry-ing upload forever. the logic makes sense for "transient" errors (e.g. network/service failure) but is potentially problematic (generating a lot of noise at minimum) for arbitrary raises.
we potentially need to investigate the class of exceptions coming from s3 library.
this feature has been requested previously e.g. https://github.com/logstash-plugins/logstash-output-s3/issues/169 and https://github.com/logstash-plugins/logstash-output-s3/issues/160 ... the later case is specific to
Errno::ENOENT
which is now handled as a non-retry case https://github.com/logstash-plugins/logstash-output-s3/pull/144