Zenodo (operated by CERN) statistics pipeline still broken since May 14th 2026
I am writing to report what appears to be a persistent failure of Zenodo's usage statistics pipeline following the service incident on May 15, 2026 (the brief incident reported between 12:25–13:11 GMT). The user-facing outage was resolved, but the statistics pipeline downstream of it appears to have stopped recording new views and downloads — and remains broken five days later.
The break point is approximately May 14, 2026. Records published from May 14 onward are showing 0 views and 0 downloads on the 'This version' counter regardless of actual traffic, and the 'All versions' cumulative counter on multi-version records has not advanced since that date.
Evidence:
Robert Koch-Institut, 'COVID-19-Hospitalisierungen in Deutschland,' Version 2026-05-15 (record ID 20196436, DOI 10.5281/zenodo.20196436): 0 views / 0 downloads on the 'This version' counter, five days after publication. The May 13 version of the same record accumulated 59 views in its one-day window before being superseded. This dataset is one of the most heavily-trafficked open datasets on Zenodo, mirrored to GitHub, indexed by OpenAIRE, and scraped continuously by automated pipelines. A reading of zero downloads over five days is not physically plausible.
The same record's 'All versions' cumulative counter has been static at 139,919 views / 113,402 downloads from the May 6 version through today's May 19 version. Six daily releases have produced zero incremental views or downloads. This is not consistent with normal traffic.
I have personally tested by clicking through to one of my own records (10.5281/zenodo.20260464) from a separate device on a different network and downloading the PDF. After 24 hours, the counter remains at 0/0 on both 'This version' and 'All versions.'
The service incident banner that previously stated 'we will provide more details in the following days' has been removed from the Zenodo homepage without the promised follow-up explanation.
Questions:
- Can you confirm the current status of the statistics aggregation pipeline?
- Is there an estimated timeline for restoration?
- If raw access logs were preserved during the affected window, is backfill of missing views/downloads planned, or will this period be permanently uncounted?
- Is there an updated post-incident report planned, given that the original banner promised one?
I would appreciate confirmation that this issue is being tracked. I have additional record IDs and timestamps available if useful for diagnostics.
Thank you for your time.
David Snider