1.6 Virtual Appliance Incident Triage
Key Takeaways
Use cluster, appliance, component process and source status to determine how widely a VA incident affects the deployment.
An Inactive VA cannot connect. A Network Error calls for checking the path to required endpoints from the affected deployment.
Storage, memory and certificate alerts identify different conditions. Compare the affected member and time window before selecting a remedy.
Validate both VA health and the original source operation after recovery. A restart confirmation only confirms that the restart was initiated.
Establish the scope of the incident
In Admin > Connections > Virtual Appliances, open the cluster card and inspect the affected member. SailPoint exposes the detail behind health badges, including timestamps and the reported error. Check VA Components for process status and the cluster’s Connections tab for attached sources. Record which source operations failed and when. An appliance can report a problem while another member still serves the source. Conversely, a healthy cluster can contain a working connector process whose target refuses authentication.
Start with the smallest scope that explains the evidence. One failing source and several successful peer sources point toward differences in target network path, connector settings or credentials. Several failing sources on the same cluster and a corresponding network error suggest a shared cluster dependency. Several clusters may share the same firewall rule or proxy. That pattern calls for checking the shared service even though the clusters and their work queues are distinct.
| Signal in SailPoint | What the signal describes | Diagnostic direction |
|---|---|---|
| Inactive State | The VA is unable to connect while in that state | Check the member’s network with the network team |
| Network Error | The cluster cannot reach required endpoints | Verify actual required destinations and egress for the deployment mode |
| Disk Warning or Disk Error | File system use above 80% or 90% respectively | Check storage and logging volume, preserve needed evidence and turn off unnecessary debugging |
| Memory Warning or Memory Error | Average memory use over the last 12 hours above 80% or 90% respectively | Compare sizing, components and workload with SailPoint’s requirements |
| Certificate Expiration Warning | A certificate has expired or will expire within 30 days | Identify the affected trust material and follow documented renewal and removal steps |
| Process Busy | The process is occupied and unavailable for new processing or updates | Check the existing work and when the process becomes available |
| Process Error | The process reported an error or could not report its status | Inspect the member, component and matching logs |
Worked incident
Two payroll aggregations fail through a production cluster during a reporting window. One cluster member reports 93% filesystem usage and shows a Disk Error. The second member is connected, and other sources on that cluster have successful operations. The first question is which member and time correlate with the failed work. Check for recent verbose logging and collect the pertinent files before making changes. If the diagnostic capture is finished, stop unnecessary debugging and follow the supported procedure for managing the log files. Confirm that file use falls and the member’s alert clears. A successful payroll aggregation on the surviving member confirms that source access works through at least one path, but it does not prove the affected member has recovered.
Now change the facts. The cluster reports Network Error immediately after an egress proxy change. Connectivity tests from both members fail, and all attached sources start failing. Compare the allow list and DNS and proxy behavior against the actual VA region and configuration. Before changing any source credentials, confirm that both VAs can reach the required endpoints again. Once health returns, rerun a controlled instance of the failed source operation and check the result. One success can confirm the tested path, while the remaining source activity may still need review.
If the error shows only an expiring certificate, capture the warning’s member and expiry details before replacing anything. SailPoint identifies the /home/sailpoint/ca-certificates/ directory in the warning’s documented remedy. After following renewal steps, check the connected source that relies on that trust material. A healthy route and correct new credentials cannot make an expired trust chain valid.
Decide whether recovery requires a restart or replacement
Use the documented operation indicated by the alert and what investigation has found. A cluster restart is available from the cluster’s menu when the situation calls for it. Confirm health after the restart instead of treating the confirmation banner as evidence of success. If the cluster cannot be restored and replacement is necessary, SailPoint describes creating a new cluster and appliances, recording the sources on the failed cluster, changing their associations, and checking connections before retiring the failed cluster. The separate-cluster case requires access to source credentials. That decision follows diagnosis and continuity planning because reconnecting many sources has an operational cost.
See SailPoint’s VA health indicators and recovery procedure and VA network requirements.
All sources assigned to one cluster begin to fail after a proxy change. The cluster reports Network Error, and cloud connectivity tests from both members fail. What is the best next technical check?
Increase the connectivity permissions of the tenant’s help desk role
Check that both VAs can reach their required cloud endpoints through the configured egress path
Reenter every source service account password
Increase the maximum account deletion percentage for each source
One member of a two-VA cluster shows a Disk Error at 93% file use while the second member still runs jobs. Which response addresses the evidence?
Remove the connector’s source authorization because other jobs still run
Assume both VAs are healthy because the cluster can still perform work
Change the source identity attributes so logs stop collecting
Inspect logging volume on the affected member, preserve relevant evidence, and turn off unnecessary debugging
Sections you finish are checked off in the contents.