Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- DNS automation requires PSC, the Cloud DNS API, and at least one automatic PSC connection.
- Check each consumer VPC and distinguish the per-instance DNS name from the global write endpoint. The latter requires Enterprise Plus and its own enablement.
- Standard reconciliation attempts to restore broken automation. Forced reconciliation can briefly interrupt existing DNS automation.
- Afterward, test DNS resolution and a new connection from the affected VPC to verify that the client reaches its expected endpoint.
Check whether PSC DNS automation is failing
Private Service Connect (PSC) gives applications an endpoint in their private network for reaching Cloud SQL. Several consumer VPCs can use the same database instance, with a different endpoint address in each network. If one VPC reports a DNS lookup failure, identify the network and name before choosing the scope of the repair.
Use authorized Cloud SQL instance details or gcloud sql instances describe INSTANCE_NAME to confirm the current configuration. A per-instance name identifies a specific instance. The primary and its replicas have separate names. A global write endpoint follows the current primary and requires Enterprise Plus with the feature and DNS automation enabled. Record which name your application uses.
Check the automation status and prerequisites
Confirm that PSC and DNS automation are enabled for the instance, the Cloud DNS API is enabled, and at least one automatic PSC connection is configured. Check the service connection policy for the affected network. Google lists a disabled Cloud DNS API, a missing policy, and backend record-creation errors as reasons the automation can fail.
Find the affected VPC in pscAutoConnections and read instanceAutoDnsStatus. If the application uses the global write endpoint, check writeEndpointAutoDnsStatus as well. Record the state and time and compare them with a VPC that still connects. For example, if the test network works but the application network fails, check the endpoint and DNS result from the failing network.
Run standard reconciliation for the affected VPC
Once you have confirmed the prerequisites, use gcloud sql instances reconcile-networking INSTANCE_NAME --psc-dns through your authorized workflow. It attempts to restore the failing parts of DNS automation. Google documents standard reconciliation as avoiding downtime for working automation, and the application still needs a post-change check.
To scope the repair to one consumer network, add --networks=projects/CONSUMER_PROJECT_ID/global/networks/CONSUMER_NETWORK_NAME. Use the project that contains the consumer VPC. If you omit --networks, Google says write endpoint DNS recreation covers all consumer networks associated with the instance. Check the instance project, consumer project, and network name before running a single-network repair.
Assess forced reconciliation if the standard run fails
Adding --force makes Cloud SQL attempt to restore all DNS automation settings and can briefly interrupt the parts that already work. Investigate why the standard run failed and record the current endpoint and DNS status. Use a team-approved change window and plan the application checks around the actual workloads.
Teams using the Cloud SQL Admin API can set force under pscDns in their reconcileNetworking request and include a networks array when needed. Set force to false for standard reconciliation or true for forced reconciliation. Confirm project permissions and notification requirements under your team's existing process.
Verify a new connection from the affected network
Check the new automation status, resolve the application’s actual hostname from the affected VPC, and compare the resulting address with that network’s PSC endpoint. Success in one VPC cannot establish success in another. For Cloud SQL Auth Proxy or a connector, verify resolution and connection from the application's own runtime environment.
Open an authorized new application or database connection and record the network, hostname, and time that worked. If DNS resolves correctly but the connection still fails, check the network path, database identity and permissions, and application settings. If a custom hostname still fails to resolve, check its CNAME and target record.
Sources
Frequently Asked Questions
Will reconciliation repair a manually maintained private DNS record?
This command addresses PSC DNS automation. Check manually maintained zones, records, and custom CNAMEs separately, including whether the application’s VPC can see the records.
Which project ID belongs in --networks?
Use the project that hosts the consumer VPC. Select the instance name in the Cloud SQL instance’s project context. These can be different projects.