Skip to content
Notifications
Clear all

Anyone else having issues with the cloud connector dropping sessions?

1 Posts
1 Users
0 Reactions
71 Views
(@amandaj)
Honorable Member
Joined: 4 months ago
Posts: 516
Topic starter   [#19702]

I am conducting a detailed analysis of user session continuity for a critical checkout funnel, and my instrumentation has identified a recurring point of failure: the Radware Cloud Connector appears to be dropping sessions intermittently. This is manifesting as a statistically significant increase in session-based errors (specifically, `ERR_CONNECTION_RESET` and timeouts) during multi-step processes that should be handled within a single, persistent session.

My environment and observed symptoms are as follows:
* **Deployment Model:** Hybrid, with on-premises applications routed through the Radware Cloud Connector for DDoS protection and load balancing.
* **Affected Workflows:** User registration (3-step form), shopping cart checkout (5-step process), and any AJAX-heavy dashboard interactions that rely on persistent WebSocket connections.
* **Observed Pattern:** Failures are not consistent. They occur in approximately 1.2% of all sessions, but within affected sessions, the failure rate for subsequent requests after the initial drop jumps to over 85%. Correlating our application logs with Radware's event logs shows a cluster of `Session Terminated` events coinciding with our client-side error tracking.

I have ruled out application-layer issues through isolated testing. The problem only manifests when traffic is routed through the Cloud Connector. My current configuration for session persistence is standard:

```json
{
"persistence": {
"method": "COOKIE_INSERT",
"timeout": 3600,
"cookie_name": "RS_STICKY"
},
"health_check": {
"protocol": "HTTPS",
"send_interval": 30,
"response_timeout": 5
}
}
```

My primary questions for the community are:

1. Has anyone else observed a correlation between specific traffic patterns (e.g., rapid successive POST requests, long-polling connections) and session drops via the Cloud Connector?
2. Are there known limitations regarding idle timeout versus absolute session timeout that might be triggered by multi-step user processes where delays between steps are user-driven (sometimes 2-3 minutes)?
3. Beyond the basic cookie persistence, are there more granular configuration parameters—perhaps related to TCP buffer sizes or maximum concurrent requests per session—that you have adjusted to stabilize similar workflows?

I am particularly interested in any diagnostic steps beyond the standard admin console. For instance, comparing packet captures before and after the connector for an affected session would be my next course of action, but I am seeking to validate if this is a known issue with a documented mitigation strategy.

Any shared experiences or configuration insights would be invaluable for my ongoing root-cause analysis.

— Amanda


Data > opinions


   
Quote