Browse and search documentation
Troubleshooting and support
Investigate sign-in, onboarding, job and notification problems and prepare useful redacted information for support.
On this page
Preserve useful context first
- Record time and timezone, page, expected result and actual symptom. Note the last known good time and recent upgrades or configuration changes.
- Record platform and Agent versions, deployment type, OS and scope. Distinguish all-user, single-project and single-target failures.
- Preserve the original execution or session record. Query an unknown state before submitting more operations.
Check by symptom
| Symptom | Check first | Avoid |
|---|---|---|
| Cannot open after installation | Installer health checks, actual address, network entry point and HTTPS | Using 127.0.0.1 as a remote server address |
| Sign-in or MFA failure | Tenant, identity provider, account state and device time | Disabling MFA or sharing administrator passwords |
| Empty asset list | Project access, filters and synchronization results | Assuming resources were deleted |
| Agent offline | Service, last heartbeat, gateway networking and certificates | Repeatedly deleting and re-enrolling |
| Job timeout or partial failure | Per-target state, failed stage and partial effects | Rerunning on every host without checking results |
| Notification not received | Actual sending configuration, channel, silences, queue and delivery records | Treating a recorded message as proof a person received it |
| AI analysis unavailable | Connection mode, model connectivity, credentials, knowledge retrieval, and whether the License page lists ai among licensed features | Relabeling unknown results as healthy or repaired |
Platform diagnostic bundle
If your Compose delivery includes the diagnostic tool, an authorized administrator can run it from the delivery directory. It collects platform host and container information read-only, not diagnostics for Agent-managed target hosts.
- The bundle includes host resources, container health and exit state, recent redacted logs, database health metadata and volume usage. Environment variable names, not values, are collected.
- By default it writes chronoops-diag-<timestamp>.tar.gz under /tmp with 0600 permissions. Inspect the manifest and contents before sending; automated redaction does not replace review.
- It does not include business-table data, recordings or full job outputs. Use a deployment-appropriate support procedure for Helm environments.
bash scripts/delivery/diagnostics/chronoops-diagnostic-bundle.shSend an actionable support request
Contact weiwendi@aiops.red with a summary and necessary scope first. Do not attach production logs, database exports or recordings to ordinary email. Agree authorization and a transfer method before sharing additional material.
Version / 版本:
Deployment / 部署方式:
Time and timezone / 时间与时区:
Affected feature / 功能页面:
Expected result / 预期结果:
Observed result / 实际现象:
Reproduction steps / 复现步骤:
Recent changes / 最近变更:
Redacted error summary / 脱敏错误摘要:When to stop and ask
Stop and contact the responsible owner if target identity differs, signature verification fails, a host fingerprint changes, credentials may be exposed or a restore would overwrite existing data. Do not bypass checks, delete records or expand scope merely to make a workflow proceed.
Send your deployment version, page and a redacted description so we can investigate and update the guide.
weiwendi@aiops.red