2024-05-28 10:45 UTC - We are investigation slow responses from queue API.
2024-05-28 11:25 UTC - We have identified the root cause. All services are now operating normally, but some jobs may be delayed.
2024-05-23 12:30 UTC - We are investigation slow responses in jobs list requests that affects the UI responsiveness.
2024-05-23 13:48 UTC - We have identified the root cause of the overload and took action to address it. The platform is stable at the moment. We continue to monitor it and are working on a permanent solution.
2024-05-20 05:53 UTC - We are investigating job delays in AWS us-east-1 stack (connection.keboola.com). The next update will be provided when we have more information.
2024-05-20 06:38 UTC - We have identified the root cause of the overload and took action to address it. The platform is stable at the moment. We continue to monitor it and are working on a permanent solution.
2024-05-19 16:24 UTC - We are investigating the higher error rate and job delays in AWS us-east-1 stack (connection.keboola.com). The next update will be provided when we have more information.
2024-05-19 16:49 UTC - The platform has stabilized, and the job backlog is being cleared. Jobs were delayed but not errored as a result of this incident.
2024-05-15 12:14 UTC We are experiencing Facebook ads extractor component to fail due to Facebook outage. Facebook reports the outage is already solved, however we are still sill seeing components failing on "Service temporarily unavailable", which is error message directly from Facebook api. We continue monitoring the issue and wait for Facebook to resolve the outage correctly. More info on https://metastatus.com/ or https://developers.facebook.com/community/threads/3832997983612514/
2024-05-12 We have experienced some jobs stuck to continue and failed between 03:00 - 03:45 UTC. The root cause was invalid job node, we had to shut it down to prevent further issues.
Since 2024-05-06 21:30 UTC, the dbt Cloud Job Trigger component has been failing due to the error 'Missing columns: scribe_enabled.' We are currently investigating the root cause and will provide more information as soon as it becomes available.
UPDATE 2024-05-07 08:43 UTC - Since 08:18 UTC, we have not observed any further errors. The DBT Cloud has resumed returning the scribe_enabled field, which has resolved the failures. We are currently investigating the cause of this change on their side.
UPDATE 2024-05-07 11:08 UTC - We have continued to observe a few occurrences of errors. A new version of the component was deployed at 10:05 UTC, designed not to fail if the scribe_enabled field is missing in the response. In such cases, scribe_enabled will be set as empty.
We have not observed any further errors since the deployment of the new version. The incident is now resolved.
Due to change how we allow projects manage their tokens, projects using KBC Project Metadata component (kds-team.ex-kbc-project-metadata-v2) are failing while requesting token for a given project.
This component is however experimental and not longer maintained, so we're investigating best scenarios to mitigate failing jobs in affected projects. For now, we're reverting the change. It should propagate within ~90 minutes from this post.
Thanks for understanding.
All DBT and No-code transformations have been affected by a bug and no tables were imported from these transformations to Storage (in process of output mapping). This issue has been detected, root cause identified and version with this issue was rollbacked.
DBT and no-code transforamtions running in interval 9AM CEST 2024-05-02 - 2pm CEST 2024-05-03 might be affected, and re-run is recommended.
This issue affected https://connection.north-europe.azure.keboola.com/ only.
Today since 12:00 UTC we suffer jobs as well as sync actions failing to start due to bug introduced in the last deploy. We are now reverting to the previous working version.
Update 13:15 UTC (resolved) : We have successfully reverted backend to the previous working version and platform is back operational now.
Update 13:30 UTC: Upon further investigation and damage analysis, we found that Queue v2 jobs were delayed in starting because of this incident. That could result in delays of up to one hour.