Databento Outage History

Blue Ocean ATS (OCEA.MEMOIR) suffered a gap between 20:41~20:48 on the A-side feed. Recovering from the B-side is not possible due to an outage with TNS. This impacts all schemas but especially order book schemas such as MBO, MBP_1, MBP-10, BBO-1s, and BBO-1m. Records in these schemas will be flagged with F_MAYBE_BAD_BOOK indicating the book history isn't complete for the rest of the session.

We've remediated the underlying issue with batch job processing, the system has been catching up on the backlog of jobs to process. Jobs will resume processing at a normal rate within an hour.

Intraday data for OPRA.PILLAR is currently being published with a delay. We have restarted the affected pipeline and expect it to catch up. Live data is not affected.

We are continuing to work on a fix for this issue.

Our primary link between NY4 and our primary datacenter went down, reducing the available bandwidth needed for us to upload and process the OPRA.PILLAR for our historical API. In working with remote hands and our carrier we've identified that a hardware port failed on a network router in our BO1 datacenter. After migrating the physical connectivity we have restored the link. This only impacts accessing today's data (Friday 2026-09-04) for OPRA.PILLAR through `timeseries.get_range` or `batch.submit_job`. The Live APIs are not impacted. No data loss has occurred, processing will resume as usual on the next trading day.

The authentication issues are resolved, so clients connecting over internet should see service restored. We are continuing to investigate the BGP issues affecting customers over dedicated connectivity.

Our GLBX.MDP3 (CME) dataset after 2026-08-13 11:30:00 UTC is currently delayed. The captured data is not lost, and will be available later today. Live data is not impacted.

Our system is currently catching up with processing intraday GLBX.MDP3 data.

The issue has been identified and a fix is being implemented.

We are continuing to work on a fix for this issue.

Our batch job processing system fell behind, processing is resuming with the backlog of jobs. (There is no need to re-submit any jobs.) We're expecting the system to be caught up by 10:00 UTC.

The fixes implemented earlier helped the situation temporarily, but the batch system has fallen behind again. This is impacting the rate at which customer jobs are being completed. We are working on a solution.

We're investigating an issue with an internal DNS server, which is causing intermittent API 500 errors. We expect this to be resolved within the next 15 minutes.

The issue has been identified and a fix is being implemented.

There were unrecovered gap in one the XNAS.ITCH gateways. We've removed the gateway from the pool. Customers are instructed to reconnect where they'll be connected to a different gateway

We're aware of delays publishing intraday historical data for CME Globex (GLBX.MPD3). This is related to the storage outage from earlier today.

Historical data for GLBX.MDP3 is continuing to catch up. Data from 2026-06-10 has been made available (T-1 complete).

We're investigating an issue with high latency in one of our internal storage systems. This may impact performance of our Portal and Historical APIs, and lead to timeouts. The Live APIs are not impacted.

One of our email service providers is experiencing delays. This includes emails as part of the signup process, team invites, and budget alerts. Support and sales emails are not impacted: you can still reach us through any of the options on https://databento.com/support

We're investigating issues with our historical API, users may see timeouts or HTTP 504 errors while this issue is ongoing.

XNAS.ITCH data after 2026-05-19 is temporarily unavailable while we remediate the issue

The API timeouts have been mitigated, however XNAS.ITCH data for 2026-05-19 (Tuesday) and 2026-05-20 (today) is temporarily unavailable.

We have identified an issue with our batch job storage system. Batch jobs are currently unavailable to download, and are not being processed. Jobs can still be submitted, but they will be queued up to process until the incident is resolved.