Known Issues for Cloud Run

Stay organized with collections Save and categorize content based on your preferences.

This page lists known issues for Cloud Run.

You can also check for existing issues or open new issues in the public issue trackers.

For help troubleshooting common issues encountered with Cloud Run, see troubleshooting.

Google Cloud services not yet supported

The following table lists services that are not yet supported by Cloud Run.

Service Notes
Web Security Scanner

High request latency with custom domains when invoking from some regions

Requests to Cloud Run services using custom domains can have a very high latency from some locations. This issue is more pronounced for Cloud Run services in asia-northeast1 and us-east4. If you observe this issue, you can achieve greater performance with Cloud Load Balancing using a serverless NEG.

HTTP/2 Push is not supported

Cloud Run supports HTTP/2, but not HTTP/2 Server Push.

Using HTTP/2 with IAP is not supported

Cloud Run services with HTTP/2 enabled encounter an infinite redirect loop upon request when secured with IAP. This behavior is not intended and Google anticipates addressing this in the future. For now, to avoid this behavior Google recommends using the default HTTP/2 configuration (disabled) when a service sits behind IAP.

Using VPC with Cloud SQL with egress settings

When using the Public IP method of connecting to Cloud SQL, you cannot use a Cloud SQL connection in Cloud Run with a serverless VPC connector that has the egress setting all-traffic.

Unsupported HTTP request methods

Requests with HTTP methods TRACE and CONNECT are not supported and therefore will not be received by services running on Cloud Run.

Reserved URL paths

It is not possible to use the following URL paths:

  • /eventlog
  • Paths starting with /_ah/
  • Some paths ending with z

Reduced availability when using 3 or less maximum instances

Revisions using a maximum number of instances of 3 or less might experience unexpected downtime.

Multi Architecture Image Format is not supported

The Multi Architecture Image is not supported at this time.

HOME environment variable is set to /home in first generation execution environment

The Cloud Run first generation execution environment sets the HOME environment variable to /home. The second generation execution environment does not set this variable.

You can't mount secrets on /tmp in first generation execution environment

If you use first generation execution environment, you cannot mount secrets on /tmp directly. However, you can mount secrets on any subdirectory of /tmp.

Second generation execution environment overrides /tmp

The second generation execution environment mounts an empty tmpfs to /tmp. This means that any data in your container's /tmp is inaccessible to your service.

Cloud Run job tasks may be spuriously marked as retried

In some situations, a task can be marked as having been retried, when in fact the task succeeded on its first try. Until this issue is resolved, Google recommends keeping the --max-retries parameter set to 3 or higher to avoid spurious execution failures.

IPV6 doesn't work with vpc-egress set to all-traffic

For Cloud Run jobs and any Cloud Run service using the second generation runtime environment, IPV6 doesn't work if vpc-egress is set to all-traffic.

Reduced availability for containers with a startup time longer than 10 seconds

Requests waiting for a container instance to start are kept in a queue for a maximum of 10 seconds. As a consequence, if a container has a startup time longer than 10 seconds, requests sent to it will be aborted. This can happen after a deployment that receives 100% of the traffic or when requests need to be sent to a new instance during autoscaling.

You can mitigate this issue by gradually rolling out revisions and by optimizing startup time.

503 errors from very high request rates at the container level

Under rare circumstances during heavy traffic and rapid scaling, connection rates to a given container using HTTP 1 can result in exhaustion of available TCP sockets, due to sockets remaining in a TCP TIME_WAIT state. This only occurs when a single container receives requests at a rate greater than 800 per second for a sustained period (at least one minute) which is uncommon. This can result in Cloud Run returning 500 errors from the service for a small percentage of requests. Services using HTTP/2 are able to reuse connections and avoid this issue.

Cloud Run jobs quota issues

Some customers may experience quota issues in Cloud Run jobs, which are in Preview. A fix is currently being worked on. In the meantime, if you run into this issue, try requesting a quota increase to the Job run requests per minute per region quota for your region.

Elevated job task failures

Cloud Run jobs (Preview) are experiencing higher failure rates, primarily affecting jobs that have large numbers of tasks. A fix is currently being worked on. In the meantime, try increasing the number of retries to the maximum value of 10.

Jobs running longer than usual

Cloud Run jobs (Preview) with high task counts are currently experiencing longer or unpredictable execution times. A fix is currently being worked on.