Is Treasure Data down?

Between 7/25 and today, 7/27 at 11:30 AM JST, we experienced a performance degradation issue affecting BulkLoad (Data Connector Import), Result Export (Activation) and Bulk Import. Following reports received this morning, our engineering team investigated the issue and identified high system load under specific conditions (increasing delay in internal components communication) as the root cause. The issue has been resolved by full rotation of our server nodes, and full performance was restored as of 11:30 AM JST. We sincerely apologize for any inconvenience caused. We are currently reviewing our monitoring and alerting rules to prevent similar issues in the future.

We are observing degraded performance in Treasure Work and tdx in the Tokyo region. This issue is limited to the Claude Haiku 4.5 model in the Tokyo region only. Users may notice that requests using Claude Haiku 4.5 return errors or time out. At this time, we do not see any issues with other models. As a workaround, users can select a different model. We are actively investigating this issue and will aim to provide an update in 30 minutes.

We are continuing to investigate this issue and will provide another update in 30 minutes.

We are continuing to explore this issue and are currently evaluating a potential fix. We will update as soon as our testing is complete.

We are continuing to work through a potential fix and working to identify any potential contributing factors. Again, as the issue is localized to Haiku 4.5 models in the Tokyo region, we recommend users temporarily use a different model if they are experiencing any elevated error rates. We apologize for the inconvenience and will have another update soon.

A fix has been implemented and we are monitoring the results.

This issue has been resolved. Treasure Work and tdx are operating normally, and the Claude Haiku 4.5 model in the Tokyo region is no longer experiencing errors or timeouts. Thank you for your patience.

We are experiencing Hive query performance degradation in the US region. This issue began on June 25, 2026. Users may observe longer than expected query execution times, particularly for long-running jobs with large data volumes. We have identified the cause and are working on a fix. We will provide an update in 30 minutes.

We are experiencing Hive query performance degradation in the US region. This issue began on June 25, 2026. Users may observe longer than expected query execution times, particularly for long-running jobs with large data volumes. We have identified the cause. To address this, we are validating a cluster configuration change prior to applying it to the production environment. We will provide an update in 30 minutes.

We are continuing to work on a fix. Our team is validating a cluster configuration change prior to applying it to the production environment. We will provide an update in 30 minutes.

We have provisioned a new cluster with sufficient capacity and are now routing jobs to it. Query performance is expected to recover gradually as the cluster stabilizes. We are monitoring the situation closely. We will provide an update in 1 hour.

Hive query performance degradation in the US region has been resolved. We have successfully provisioned a new cluster and routed all traffic to it. Systems are now operating normally. Thank you for your patience.

Resolved — Some result export (data export) jobs in the US, Tokyo, and EU regions may have failed or been delayed between June 24, 2026 21:00 UTC and June 25, 2026 05:26 UTC. During this period, underlying compute nodes were intermittently unstable, which could cause export jobs to terminate before completing. In some cases an affected job may have ended without a clear error message.The issue has been resolved and all systems are operating normally. Data integrity is not impacted — source data in Treasure Data was unaffected; only the delivery of export jobs to external destinations during this window could have been disrupted.If you ran a result/data export during this window, we recommend confirming it completed and re-running it if the output looks incomplete. We will also contact affected customers directly. Thank you for your patience.

We identified a delay in recording email event data (such as open, click, bounce, delivery events, and error events) to customer event tables. Affected period (approximate): - June 22, 03:45 UTC – June 23, 04:15 UTC - June 23, 07:00 UTC – June 24, 07:30 UTC During these periods, email event data recording was delayed. Email delivery itself was not affected — emails were sent and received normally. The issue has been resolved and all systems are operating normally. Delayed events have been fully processed. If you ran queries on event or error_event tables during the affected periods, we suggest rerunning those queries to get accurate results. We apologize for any inconvenience.