ImpulseValkey
Troubleshooting
Common situations across provisioning, connectivity, and lifecycle — and where to look. When in doubt, start at Diagnostics → Health Checks and run every check: together they isolate provider, DNS/TLS, server, cron, schema, and sync problems without touching a live instance.
1. Start at Health Checks
Section titled “1. Start at Health Checks”Open Diagnostics → Health Checks and run the checks. Each check is read-only and reports pass, fail, or skip:
| Check | Confirms |
|---|---|
| Cron Jobs | Core registration and the Valkey provisioning, usage, snapshot, and credential-rotation tasks are registered. |
| VPS Providers | Every configured provider API responds, without needing SSH. |
| DNS & TLS Settings | Base domain, Cloudflare token, zone id, ACME inputs, and TLS port are valid. |
| Valkey VPS Servers | Each online server’s Admin Proxy health endpoint answers; refreshes the health timestamp. |
| Snapshots | The snapshot target preference and operator S3 fallback settings resolve. |
| Email Delivery | ImpulseCore’s mailer and this module’s sending provider are configured (no test mail is sent). |
| WHMCS Product Sync | Synced plans still point at existing WHMCS products using the module. |
| Cloud-Init Template | The bootstrap template exists, has the required placeholders, and installs the Admin Proxy. |
| Schema & Runtime | Expected tables, module version, PHP, database, and loader state. |
| Provisioning Preflight | A no-resource readiness check for dedicated provisioning. |
2. Provisioning
Section titled “2. Provisioning”Order stuck pending
Section titled “Order stuck pending”Dedicated builds run on the 2-minute provisioning cron. Confirm Core’s cron runner is firing (Health Checks → Cron Jobs) and that the provider API responds (Health Checks → VPS Providers). A build that makes no progress for the provision timeout (15 minutes by default, under Settings → General & Access) shows as provision_failed; retry it from the server’s row, or run recovery server actions on it.
Shared order fails at Create
Section titled “Shared order fails at Create”No shared pool box has a free slot in the requested region. Provision or free capacity on a pool box, or remove the explicit region constraint. Region requests are a hard placement constraint — an order naming a region with no capacity returns a specific no-capacity error rather than placing elsewhere.
3. Server & connectivity
Section titled “3. Server & connectivity”Server unreachable
Section titled “Server unreachable”Run a Health Check server action and Fetch Bootstrap Log. Check the Admin Proxy URL on the server row, and that the server’s firewall still admits this WHMCS server: it opens the Admin Proxy ports (9098 for HTTPS, plus the Admin Proxy port setting, 9099 by default) only to the IPs set under Settings → General & Access.
TLS warnings
Section titled “TLS warnings”The certificate on a server’s Valkey port is self-signed by design, so the TLS cert is self-signed alert is expected on every dedicated server — switch that alert off under Settings → Alerts & Notifications if you don’t want it. Reissue TLS renews the Admin Proxy’s certificate, not the Valkey port’s. Diagnostics → Health Checks also has a TLS certificates panel showing what each server presents.
Customer can’t connect
Section titled “Customer can’t connect”Check four things: the client is using TLS / the rediss:// scheme (not plain redis://); the client trusts the instance’s self-signed certificate (see Connecting a cache client); the current password is in use; and, if an IP allowlist is set, the application server’s IP is on it. On a shared plan, the port is the one in the instance’s own connection string, not 6380.
4. Credentials & metrics
Section titled “4. Credentials & metrics”Customer can’t connect after rotation
Section titled “Customer can’t connect after rotation”The old password is valid only during the grace window (default 30 minutes). Have the customer re-reveal the connection string from the Connect & Access tab and update their application to the current password. A credential-rotation cron (every 5 minutes) revokes expired grace passwords on the server. A plan change on a dedicated service rotates the password the same way.
Metrics look stale
Section titled “Metrics look stale”Metrics come from the hourly usage cron. Use the Refresh Metrics server action (or the service-page Refresh Metrics button) to sample on demand.
5. Licensing
Section titled “5. Licensing”License prompt won’t clear
Section titled “License prompt won’t clear”Confirm a valid IVKEY- key is entered under ImpulseCore → Modules & Licensing → Licenses, then check Diagnostics → License for status and grace. Licensing is centralised in Core, not entered in the ImpulseValkey addon.
6. When in doubt
Section titled “6. When in doubt”Start at Diagnostics → Health Checks and run every check — together they isolate provider, DNS/TLS, server, cron, schema, and sync problems without touching a live instance. Diagnostics → Activity shows the audit trail for any action that looks wrong, and Diagnostics → Cloud-Init Templates lets you confirm the bootstrap payload has the required placeholders and installs the Admin Proxy.
Still stuck? Contact Impulse Software support — the Help Center at support.impulsesoftware.com, or email [email protected]. Diagnostics → Help generates a masked diagnostic report to include.