ChangesPrismaPrisma ORM engine binary download issues causing Accelerate and Prisma Postgres errors
Prisma: Major incident: Prisma ORM engine binary download issues causing Accelerate and Prisma Postgres errors
Summary
Summary pending. The diff below shows exactly what changed.
Incident page: the full timeline of updates
What changed
Struck in red is what Prisma removed; blue is what they added. Muted text is unchanged context.
Before: not present
After: Prisma ORM engine binary download issues causing Accelerate and Prisma Postgres errors
Before: not present
After: major
Before: not present
After: resolved
Before: not present
After: 2025-07-18 18:00 UTC
Before: not present
After: 2025-07-18 23:37 UTC
Before: not present
After: Prisma Accelerate, Prisma Postgres
Before: not present
After: Description: We have identified and mitigated an issue with Accelerate, also affecting Prisma Postgres, where Prisma ORM binaries are unable to be downloaded. The root cause has been found and fixed and we are monitoring Accelerate and Prisma Postgres.
Affected users: All Accelerate and Prisma Postgres users, and a subset of Prisma ORM users
Root cause: Identified and fixed
Before: not present
After: As we continue to monitor we have an update on the root cause and resolution of this incident:
- When a new version of the ORM is published it uploads a binary to R2 and a failover to copy to S3
- ORM users, including Accelerate and Prisma Postgres, require these binaries when starting up query engines
- The failover to S3 triggered but ORM, Accelerate, and Prisma Postgres were not able to download binaries
- We resolved the incident by overriding the failover and pointing to R2
We're investigating both the cause of the failover the issue related to downloading binaries from S3.
Before: not present
After: We are currently investigating this issue.
Before: not present
After: The issue has officially been resolved. After identifying and resolving the root cause we experienced issues while restoring Accelerate to service.
We're continuing to monitor closely and will provide another update with more details on what delayed bringing Accelerate back to service.
Before: not present
After: After resolving the original Prisma ORM binaries download issue we observed many users attempting to create query engines at the same time. Accelerate keeps a pool of warm instances to serve these requests quickly and when that warm pool is depleted we fallback on creating cold instances.
We identified a bug in the cold instance path which prevented us from bringing Accelerate online for all users fast enough. While some users were able to get instances from the warm path as new instances were added to the pool, it wasn't fast enough to serve the backlog of users who were requesting new instances. Once we found and patched the bug Accelerate was able to serve all instance requests within minutes.
Because this only affected the creation of new instances, users with instances already running would have not experienced any issues.
Thank you for your patience. We're already working on collecting data and insights for our postmortem which always aim to make our products and services better.
Before: not present
After: The number of active instances and HTTP traffic has returned to typical levels, warm pools are healthy and serving new instance requests, and errors are down.
While systems are now operational and this incident has been resolved, we will continue to monitor things closely.
First version of this incident archived by Rewindex. View the live incident (opens in new tab).
History of this incident
Friday, July 18, 2025
1 incident change
- MajorMajor incident
Prisma ORM engine binary download issues causing Accelerate and Prisma Postgres errorsthis change