Overview
A public sector agency runs OpenSearch for case search and audit logging under strict availability and access-control requirements. AceMQ provides continuous support covering cluster health, security plugin configuration, and version upgrades across 24/7 coverage.
Challenge
The agency's operations team was competent but thin, and OpenSearch incidents tended to arrive at inconvenient hours. Fine-grained access control through the security plugin meant that a misconfigured role mapping could silently hide documents from an entire user group, which is an availability problem that does not show up in cluster health. Upgrades were repeatedly deferred because nobody wanted to be the one to break document-level security.
Environment
Self-managed OpenSearch on-premises with fine-grained access control, document-level and field-level security, air-gapped patch distribution.
Approach
Support covers both the infrastructure layer and the security plugin configuration that most vendors treat as the customer's problem. We hold role mapping and tenant definitions under review, validate them against test identities before changes reach production, and run upgrades as rehearsed procedures on a staging cluster first.
Solution
- 24/7 coverage with a 15-minute emergency SLA and direct access to senior engineers familiar with the agency's topology
- Security plugin role mapping, document-level, and field-level security reviewed and regression-tested against representative identities
- Version upgrade planning including plugin compatibility checks and rehearsal on a mirrored staging cluster
- Cluster health, shard allocation, and JVM monitoring reviewed on a recurring cadence rather than only during incidents
- Patch and CVE guidance adapted to an air-gapped distribution process
- Incident post-mortems delivered with concrete configuration changes rather than generic recommendations
Outcome
The agency has stayed current on OpenSearch versions instead of falling years behind, and access-control regressions are now caught in staging. Out-of-hours incidents reach a senior engineer immediately rather than escalating through a queue.
Technologies
OpenSearchOpenSearch Dashboards
Linux
Kubernetes
Amazon S3
Terraform
Prometheus
