What happened: Between 16:42 and 07:38 CEST, charging control decisions for cloud connected EVs were temporarily not executed. Affected vehicles remained in their last known state for the duration of the incident.

Impact: Approximately 5% more charging sessions missed their target compared to baseline. The majority of charging sessions completed successfully. No data was lost, and all systems resumed normal operation automatically after the issue was resolved.

Resolution: A fix was deployed and normal operation was restored by 07:38 CEST on August 12.

The maintenance work has been completed successfully.

There won't be any interruptions or rejections for the Webhook Subscription API anymore.

On 2026-07-22, a coordinated release across the customer-facing frontends caused widespread 401 authentication errors for users. The issue stemmed from a change in the authentication flow and an audience configuration change. The backend’s newly tightened validation logic rejected these tokens, causing users to be unable to log in or maintain their sessions.

Impact

Between 09:30 and 10:15 CEST, users attempting to authenticate were partially met with 401 errors. This disrupted normal frontend operations. While the primary issue was mitigated within 45 minutes, residual errors persisted for several hours due to browser-side caching of the previous frontend version.

Root Cause

The incident was caused by two intersecting factors:
* Backend validation: The backend was updated to implement stricter validation on JWT audience claims.
* Frontend refresh logic: The frontends' token refresh flow - when requesting a new access token - was returning multiple audiences in a way that did not match the strict expectations set by the backend.

The configuration change moved the frontends to a new audience, which triggered this mismatch, whereas the previous configuration had relied on legacy behavior that the backend was no longer implicitly supporting.

Detection

The incident was detected by automated alerts which fired shortly after the v5.13.0 rollout.

Resolution

The following steps were taken to mitigate and resolve the incident:
* Rollback: The frontend changes were rolled back to the previous stable versions for both frontends.
* Backend Configuration: The backend validation logic was amended to temporarily allow the legacy audience value, ensuring that existing sessions remained valid while the frontend could be safely updated.

The Flex/ToU Optimization on C2C connected EVs was fully restored by the fix deployed earlier.
This incident is resolved.

Time-of-Use (ToU) optimization has been fully restored and is running normally across all systems as of today, Monday, June 1st, at 15:15 CEST.

The disruption originally began on Friday, May 29th at 12:15 CEST, initially affecting a subset of approximately 1,500 systems. Between 12:15 CEST and 16:15 CEST on Friday, these specific systems applied sub-optimal decisions due to the standard 4-hour decision window, after which all ToU decisions ceased for them.

The impact expanded to all remaining systems today, Monday, June 1st. Between 10:00 CEST and 14:00 CEST, the entire fleet applied sub-optimal decisions because no new ToU decisions could be generated, forcing systems to rely on the remaining 4-hour window of previously created decisions. Once those older decisions expired, no ToU decisions were applied across any systems between 14:00 CEST and 15:15 CEST.

The root cause of the incident was an internal deployment on our side, which was subsequently exacerbated by a gridbox release that occurred this morning. The system was successfully stabilized today, Monday, June 1st, at 15:15 CEST, and correct ToU decisions are once again being actively applied everywhere.