Dust had 43 outages in the last 2 years totaling 9h 49m of downtime — averaging 1.8 incidents per month.
There were 43 Dust outages since December 2, 2024 totaling 9h 49m of downtime. Each is summarised below — incident details, duration, and resolution information.
From 10:12 UTC to 10:45 UTC, users that were not editors of an agent lost ability to chat with it or see its details. This was due to a faulty logic change that has now been fully understood and fixed.
Dust is currently unavailable in the EU region due to an expired certificate. Our engineering team is working to resolve this issue as quickly as possible.
New certificates are being provisioned. Service restoration is expected between 15 to 30 minutes.
Dust EU is back online with proper certificates.
We are currently investigating this issue.
We have identified the problem and have rolled out a fix. We are monitoring the situation.
Incident is fully resolved
During that time, in a new conversation, uploading a file from one's computer would fail. This did not affect existing conversations, or attaching from datasources. This was due to a faulty deploy; the problem has now been fixed.
We are experiencing a critical issue affecting data source connections across all workspaces. We are investigating the issue and hope to have it resolved as soon as possible. During this time, agents may not have access to their configured knowledge bases, which could result in incomplete or inaccurate responses.
We are experiencing a critical issue affecting data source connections across all workspaces. We are investigating the issue and hope to have it resolved as soon as possible. During this time, agents may not have access to their configured knowledge bases, which could result in incomplete or inaccurate responses.
The issue has been identified and we are working on a fix
We are experiencing a critical issue affecting data source connections across all workspaces. We have identified the issue and are preparing a fix. During this time, agents may not have access to their configured knowledge bases, which could result in incomplete or inaccurate responses, and agent creation and modification is disabled.
We are experiencing a critical issue affecting agents using data sources. Agents may not have access to their configured knowledge bases, which could result in incomplete or inaccurate responses, and agent creation and modification is disabled. We are rolling out a fix, the issue should be resolved in 15 minutes.
The fix has been rolled out. Agents should now answer properly, creation and edition of agents is available again.
Authentication is currently not possible on Dust due to the failure of one of our providers. They have identified the issue and are actively working on a fix. You can follow their progress on https://status.auth0.com
Auth0 seems to have resolved the issue. We will continue monitoring.
As part of a change to enforce stricter security policies (CORS), use of extension was made unavailable during that time. This has now been fixed.
All agents are currently not able to answer and will throw an error. We are actively working on a resolution.
:large_green_circle: Resolved: Agents not responding Between 10:15am UTC and 10:40am today, a restrictive modification to our CORS policy caused disruptions. The change has been rolled back, and all agents are now operational. You can continue using Dust as usual.
Agents are currently unavailable. We're working on the situation and it should be resolved soon.
Incident resolved, agents are back to being fully functional.
Website synchronization was not active since Jan 22, 2025. This was due to an erroneous cancellation of the synchronization scheduler. As a result, data on websites were stale until that date. Websites would still be crawled when added, or when manually resynced. Synchronization has now resumed and is back to normal
Due to limits on the Google Drive API, Google Drive synchronizations might encounter delays. We have found and corrected the root cause. The incident is over
This incident has been resolved.
From 13:22 UTC to 14:58 UTC, conversations list in sidebar all appeared under date "Today" in a random order. This was due to a faulty migration. Issue has been fixed.
OpenAI-based assistants are slow, we are currently investigating the reason.
Since earlier today, OpenAI is experiencing errors with their Embeddings API (you can follow the situation at: https://status.openai.com), which our Search tool relies on heavily. This may cause Dust to appear slow and unreliable. We are monitoring the situation closely and will provide updates as they become available.
The OpenAI errors rate seems to have gone down drastically. Considering this incident as resolved.
Visualizations generated by models in conversations are failing to display. We are investigating the issue.
Visualizations generated by models in conversations are failing to display. We identified the issue and are rolling out a fix.
We are continuing to work on a fix for this issue.
Visualizations generated by models in conversations are failing to display. We are investigating the issue.
Visualizations generated by models in conversations are failing to display. We have identified the issue and are rolling out a fix.
Visualizations generated by models in conversations were failing to display. We identified the issue (erroneous env variable checking the origin) and deployed a fix.
We identified an higher than usual rate of errors on assistants execution likely due to networking errors in between 2 of our services. We are investigating the issue.
We identified the issue and deployed a fix. We're monitoring the situation.
Monitoring confirms incident is resolved.
Between December 11, 2024 and Jan 17th 2025, table queries on Notion databases have not been functioning due to a defective database migration on our side, which caused an ID mismatch. We have identified the issue and are in the process of deploying a fix. Only assistants using table queries on Notion databases are affected. You are receiving this message because some of your assistants have been affected.
We are actively investigating the matter and diligently working to implement a resolution. We appreciate your patience during this process.
We are continuing to investigate this issue.
We are actively investigating the matter and diligently working to implement a resolution. We appreciate your patience during this process.
We are actively investigating the matter and diligently working to implement a resolution. We appreciate your patience during this process. Default agents (@gpt4, @claude-3, @mistral, @gemini-pro) are still available.
Between 12:44pm and 1:23pm UTC custom assistants with actions became unavailable through the API and not visible on the web UI. This incident has been resolved and the service is now operating normally.
Between November 29th 6:03pm UTC and December 2nd 10:36am, Slack workflows mentioning the Dust bot were raising errors instead of answer. This has now been fixed and all Slack workflows should now proceed normally.