We had an example yesterday were a stream received unusually large amount of data

https://app.datadoghq.eu/dashboard/xnk-i3d-xj7/arcane-20-streams?fromUser=true&fullscreen_end_ts=1758801720027&fullscreen_paused=false&fullscreen_refresh_mode=sliding&fullscreen_section=overview&fullscreen_start_ts=1758628920027&fullscreen_widget=8823882781046968&refresh_mode=sliding&tpl_var_StreamId%5B0%5D=mss-eu-general-journal-account-entry&from_ts=1758626671745&to_ts=1758799471745&live=true

The stream was able to process the data, slowly merging the data into bronze, but eventually after several hours it for some reason decided to jump to current time - lookBackInterval which effective skipped all the files between 14:00-14:45

https://app.datadoghq.eu/logs?query=%40host%3Amss-eu-general-journal-account-entry-%2A%20%28%40messageTemplate%3A%22Reading%20file%20%25s%2F%25s%20from%20Azure%20Storage%20account%20%25s%22%29%20host%3Amss-eu-general-journal-account-entry-vcs4h&agg_m=count&agg_m_source=base&agg_q=host&agg_q_source=base&agg_t=count&clustering_pattern_field_path=message&cols=host%2Cservice%2C%23file&data_extractions=%5B%5B%22message%22%2C%22Reading%20file%20dataverse-prdeccocceu-unq732376bb88cd48fe9c00b159d4492%2F%25%7Bdata%3Afile%7DZ%2Fmodel.json%20from%20Azure%20Storage%20account%20daieu%22%5D%5D&event=AwAAAZl8oaY9vNrkIwAAABhBWmw4b2FhZkFBQWVBUHVCSVlIZjd3QUIAAAAkZjE5OTdjYWEtZWE2OC00ZTlhLTk1NjYtMmVhYmVkMzU2MGJkAABQoQ&fromUser=true&messageDisplay=inline&refresh_mode=paused&storage=hot&stream_sort=time%2Cdesc&top_n=10&top_o=top&viz=stream&x_missing=true&from_ts=1758718800000&to_ts=1758734460000&live=false

Does it make sense that the stream jumps, even if there are still unprocessed files ?