Voiceflow 장애 기록

We are currently investigating reports of agents failing to retrieve content from the Knowledge Base. We will provide updates as our investigation progresses.

A fix has been deployed and we are actively monitoring to ensure the issue has been fully resolved.

This incident has been resolved. Knowledge Base retrieval is now functioning normally.

There is an issue affecting the ability to upload new documents to the knowledge base.

This issue has been resolved. Knowledge base uploads are now functioning normally.

We are currently investigating an issue affecting the ability to create new projects via prompt generation. Other methods of project creation remain unaffected. We will provide updates as our investigation progresses.

We have identified the root cause of this issue and are currently deploying a fix. We will provide an update once the fix has been released.

This incident has been resolved. Project generation is operating normally.

We experienced an incident affecting secrets. A change intended to prevent secrets from being exposed in logs inadvertently broke variable substitution in certain areas of the platform. This incident is now resolved.

We experienced an incident affecting secrets. A change intended to prevent secrets from being exposed in logs inadvertently broke variable substitution in certain areas of the platform.

Multiple users are experiencing an issue where their chat widget colour settings have reverted to the default blue, despite being correctly configured.

We have identified the root cause of this issue and a fix is currently being deployed.

A fix has been deployed and we are currently monitoring its impact.

This incident has been resolved. Widget colour settings are now correctly reflecting in production as expected.

Upstream cloud provider reporting elevated errors and latency

We are aware of an ongoing AWS infrastructure issue affecting one Availability Zone in the US-EAST-1 region. Some services may experience increased latency or intermittent errors while AWS continues restoration work. Our team is actively monitoring the situation

AWS has resolved the data center issue that caused high latency and elevated errors. Our services are fully operational.

Our services are fully operational.

A subset of users are encountering the error 'Unable to load diagram, please try again.' when attempting to access their projects.

We have identified the root cause of this issue. It is affecting projects published to the staging environment. A fix is currently under review.

We have deployed a fix for this issue and are actively monitoring its impact.

This incident has been resolved. Projects are now loading as expected.

Chat widgets may be loading the development version of a project instead of the production version, even when the version ID is set to production. Our team is actively investigating.

This issue has been resolved. Widgets are now correctly loading the production version of projects as expected.

Analytics data may be slow or missing during this period. Does not affect creator, live agents or transcripts.

Transcripts affected, no new logs can currently be saved. Does not affect creator, live agents.

Transcripts logs are now being properly captured, but still issues with displaying/fetching them.

transcripts working as normal, currently recovering transcripts

We have received multiple reports of users encountering the error "Could not compile your agent, please try again later" when attempting to run or publish their project.

A fix has been deployed for this issue. We are monitoring the system to confirm stability and to identify any remaining edge cases.

The issue has been resolved and services are operating normally.

We identified an issue this morning affecting the opening of workspaces and projects. In cases where projects did open successfully, workflows were not visible. Our engineering team has identified the cause and is actively deploying a solution. We will provide an update once the fix has been fully implemented and verified. Thank you for your patience.

Services are restored.

Multiple users are experiencing issues with projects not loading in the Creator app.

A database restart has been completed and services are now operating normally. We are actively monitoring for stability.

Services are operational, and the incident is now resolved.

Multiple users report that agents are unable to retrieve information from knowledge base sources.

We have identified the root cause of the issue and are in the process of deploying a fix.

The fix has been successfully deployed and the incident is now resolved. All systems are operating normally.

Attempting to load a project shows a spinner and fails to load

We've identified the issue and are implementing a fix. Service is restored but could be degraded while it catches up.

Services are restored.

Currently, we are experiencing an issue on our platform that prevents people from opening their projects. Live agents are not impacted

The issue has been resolved. Voiceflow applied several optimizations to its internal database, and following these changes, some services temporarily stopped connecting to the transcripts database. Live agents continued to operate as expected. The only impact occurred when opening a project in the Creator tool, and some transcripts generated during this period may not have been stored in the database.

Our WAF (Web Application Firewall), hosted by Cloudflare, is experiencing an outage. [This outage is affecting millions of services across the internet](https://www.cloudflarestatus.com/). The outage means that users are currently unable to login to Voiceflow and make modifications to their projects. Agents embedded into websites and connected to phone numbers may not behave as expected due to this outage affecting upstream model providers.

Cloudflare Outage has been resolved

> Our WAF (Web Application Firewall), hosted by Cloudflare, is experiencing an outage. [This outage is affecting millions of services across the internet](https://www.cloudflarestatus.com/). The outage means that users are currently unable to login to Voiceflow and make modifications to their projects. Agents embedded into websites and connected to phone numbers may not behave as expected due to this outage affecting upstream model providers.

[The Cloudflare outage](https://www.cloudflarestatus.com/) appears to be mostly resolved, and our services are responding and stable, we will continue monitoring the situation.

The voiceflow.com website is impacted by ongoing issues at AWS, and consequently Webflow: https://status.webflow.com/.

Voiceflow.com is operational, and the Webflow incident is now resolved: https://status.webflow.com/incidents/t0q4jjjdc5qk

Some users are reporting that the analytics dashboard is not populated for October 7th, despite transcripts being created for this date.

Incident has been resolved and analytics should now reflect new interactions.

The set step is failing to save prompts to variables. The issue appears to be limited to the canvas, and agents published to production do not seem to be impacted.

The fix for the issue has been deployed. That said, affected projects will need to be recompiled, through editing the project and re-running within the canvas.