1.7 Virtual Appliance Logs and Observability
Key Takeaways
For VA connector processing, SailPoint documents the Cloud Connector Gateway log at
/home/sailpoint/log/ccg.log.Errors in
/home/sailpoint/log/otel_agent.logcan explain why an appliance’s metrics are missing from an observability backend.Correlate a source operation with the responsible cluster, candidate appliance members and UTC time window before interpreting entries.
Use temporary debug logging for a focused reproduction and confirm that the intended monitoring backend actually receives exported records.
Choose the record that answers the question
A source operation and a monitoring export can fail at the same time for different reasons. A connector log records the behavior of the connector runtime. An agent log can show that the monitoring exporter itself could not deliver telemetry. A systemd log records host service activity. An engineer should identify what they are trying to prove before collecting files, so they inspect the component involved in the failed action.
| Evidence needed | Documented location or view | Interpretation |
|---|---|---|
| Connector activity from a VA-based integration | Cloud Connector Gateway (CCG) log at /home/sailpoint/log/ccg.log | Start near the failure time and use the affected operation or source to narrow the entries |
| Errors sending appliance metrics through OpenTelemetry | /home/sailpoint/log/otel_agent.log on that VA | Inspect backend detection or configuration errors before diagnosing missing metrics as a source outage |
| Cluster or member alert and timestamp | Cluster details, followed by the relevant VA or component view | Check which member or process needs a closer look |
| VA host service logs in an external tool | Supported systemd log export configured using host-logging.yaml in /home/sailpoint | Confirm network access to the approved collector and check actual receipt |
The cluster queue can be polled by more than one VA. If the tenant’s operation view does not establish which member picked up a job, examine the relevant window across candidate members. Avoid concluding that an event never occurred solely because it is absent from one member’s local log. Also separate a SaaS connector from a VA-based one before selecting the diagnostics. SailPoint documents connector logs for supported SaaS sources through the source’s settings. Those connectors do not require a customer VA, so choosing a cluster at random cannot give their connector logs.
Reconstruct the failure
A VA-based inventory connector’s test fails at 07:42 local time and the tenant records the connector name. The incident ticket says the local clock uses a fixed UTC−05:00 offset that day. Convert the ticket timestamp to 12:42 UTC and search a short window on the cluster members. SailPoint’s standard VA time zone is UTC. Record the original date and zone in the incident evidence as well, since a conversion can cross midnight and the cluster’s scheduled aggregation time zone is a separate setting.
If the log shows that the connector rejected a response from the application, take the exact error and operation to the connector’s configuration or troubleshooting documentation. The generic failed state in the tenant is useful for finding the incident but is usually less specific than the component message. Compare timestamps to see whether an earlier connection or permission error caused a later missing-data symptom. Check the target’s own record when that service can confirm whether a call reached it.
Take a storage reading before a verbose test and keep the exact start and end times. SailPoint describes 24-hour retention regardless of file size when debugging is enabled in the cluster and replacement at 512 megabytes in its normal logging mode. A fast-growing log can therefore make a sample from before debugging hard to recover later. Save the relevant records while they are available, then correlate them with the tenant operation.
Capture and export deliberately
When normal logging gives too little information, enable documented debug logging for a controlled reproduction on the affected cluster or use the connector-specific method where applicable. SailPoint’s cluster setup documentation describes a 24-hour verbose window. Monitor storage while it is active, especially if the incident already involves disk pressure. Record the time, source, operation, and affected member alongside the diagnostic logs. Report only the fields the people receiving the evidence need.
A separate metrics example starts after changing the observability backend. Source operations still complete but the dashboard stops updating. Verify the OpenTelemetry agent configuration and inspect its own log for errors involving the endpoint or its API key. Then confirm fresh metric delivery in the destination. To validate a systemd log export, use the documented host logging configuration and check receipt of actual logs. A working metric export alone does not show that the log export was configured.
See SailPoint’s connector troubleshooting, VA observability, cluster logging settings and standard VA time zone.
A VA-based connector test fails. The team needs the local connector runtime log for the affected member. Where should it start?
The customer’s external identity provider sign-in report
The VA’s OpenTelemetry agent log
The VA’s /home/sailpoint/log/ccg.log
The source’s account schema export
An operations ticket records 07:42 at a UTC−05:00 offset. What time should the engineer use to locate that incident in standard VA logs for the same date?
12:42 UTC
02:42 UTC
07:42 UTC
11:42 UTC
Sections you finish are checked off in the contents.