Skip to content

Troubleshooting

This guide will go over common Valkey GLIDE issues. If your issues persist, please feel free to reach out to Valkey GLIDE community for help.

To help diagnose issues, you will need to enable client logging to DEBUG mode.

In many situations, clients surface errors as generic timeout errors as they try to reconnect. Enabling DEBUG logs may provide more context on the issue.

To enable logging, you will need to set the logger from within your application.

from glide import Logger, Level
Logger.set_logger_config(Level.DEBUG)

This section troubleshoots timeout errors.

During a single node failover, the entire cluster becomes unavailable. Debug logs may show CLUSTER_DOWN errors.

This may happen if your cluster only has 2 shards, which is often the setup during testing.

In a cluster with only 2 shards, losing 1 shard means 50% of hash slots are unavailable. This forces the cluster into a CLUSTER_DOWN state to preserve data integrity. GLIDE reports this as timeouts because it keeps retrying against a cluster that refuses to serve requests.

Fix: Use a minimum of 3 shards (preferably 4+) for production. With 3 shards, losing one affects only ~33% of data, allowing the remaining nodes to stay online.

Received TimeoutError on startup and failed to create GlideClusterClient.

This happens when default connection timeouts (often ~250ms) are too aggressive for VPNs, cross-region links, or high-latency networks. To fix this, you must increase the connection timeout which is separate from the request timeout.

// Node.js Example
const client = await GlideClusterClient.createClient({
addresses: [...],
advancedConfiguration: {
// Give the handshake at least 1-5 seconds
connectionTimeout: 5000
},
// This is the request timeout, NOT connection timeout
requestTimeout: 2000,
});