Return
-
Title
Quarantine due to error Cookie is Null And Maintenance Cannot Continue -
Description
In Safeguard for Privileged Passwords clusters, multiple replica appliances may enter Quarantine state during Audit Log Maintenance jobs. When this occurs, the primary appliance may lose quorum and enter a read-only state.
The following log entry is seen on the quarantined replica appliances:[E] [State.Maintenance] Cookie is null and maintenance cannot continue.In one reported scenario, one appliance had been rebuilt and re-enrolled into the cluster. Because the appliance had not previously completed the weekly audit-log purge job as a cluster member, its first purge run needed to build its cache from scratch. This caused the purge job to run for 4 hours instead of 45 minutes.The extended runtime compressed the maintenance windows for the replicas into a short period. During that time, a routine background health check changed the cluster role of each replica while the appliances were in Maintenance. This corrupted the value used to determine what state the appliance should return to after maintenance completed.When the Audit Log Purge process completed, each affected replica attempted to resume from Maintenance without a valid maintenance context. As a result, the replicas entered Quarantine. With 4 replicas in Quarantine, the primary appliance lost quorum and entered read-only mode. -
Cause
Issue in handling the SPP maintenance state during the Audit Log Maintenance. -
Resolution
STATUS:
Change Request # 729085 has been raised to the product team address this issue in a future release of SPP subject to successful QA and Product management approval.
WORKAROUND:
If the cluster is already affected and replica appliances have entered Quarantine, recover the affected appliances by rebooting the quarantined replica nodes.
To reduce the risk of recurrence:
- Avoid adding or re-enrolling cluster members immediately before the scheduled weekly Audit Log Maintenance window when possible.
- After adding or rebuilding a cluster member, monitor the next scheduled Audit Log Maintenance job closely.
- If multiple replicas enter Quarantine, reboot the affected replica appliances to restore cluster quorum.