We have since June 15. experienced a large number of incidences where our monitoring detects that one of the backends of thredds.met.no is halting due to abusive massive parallel downloads. The message on status.met.no typically becomes a "https down - Incident resolved" due to the short interval between halting the backend and restarting it. The apparent outage only affects a small number of users each time it happens and our priority users are not affected. We have tried to locate the originator of the abusive loads, but it appears that they are using cloud services where the IP address changes every 6-24 hours, so blocking is not solving it. It seems likely that the abusive loads are during data extraction from https://thredds.met.no/thredds/catalog/meps25epsarchive/catalog.html where they are accessing many YYYY/MM/DD/HH/meps_mbr_???_sfc_YYYYMMDDTHHZ.ncml at the same time. Such parallel data extractions are a breach of the Terms Of Service. We are considering other means to reduce the noise from the monitoring or blocking use of these datasets for non-priority users.
Posted Jun 18, 2026 - 09:59 CEST
This incident affected: thredds.met.no (https://thredds.met.no, MEPS ensemble).