Graph-Node Deployment and Database Connectivity Issues

Incident Report for The Graph

Resolved

A recent change in pruning behavior unintentionally triggered a spike in postgres_fdw connections, exhausting available database connections. All systems have returned to normal operation and monitoring remains in place.

Posted Apr 07, 2025 - 18:25 UTC

Update

We are continuing to investigate this issue.

Posted Apr 07, 2025 - 15:29 UTC

Investigating

We’re currently investigating recurring issues with graph-node deployments and database connectivity affecting multiple subgraphs. Users may experience failures with symptoms including "database unavailable" and "pruning failed: could not connect to server 'shard_primary'" errors, as well as delays in deploying to the studio.

Posted Apr 07, 2025 - 15:28 UTC

This incident affected: Upgrade Indexer - Miscellaneous (Deployments) and Upgrade Indexer - Subgraph Health.