Operators¶
Deploy · configure · run
Everything you need to stand up Telha and keep it healthy: install and deploy, the full configuration surface, keys and secrets, connectors, backups, retention, and day-2 operations.
-
Deployment runbook
The document of record: clean-host systemd deploy, Docker, the Windows dev/eval zip, key inventory, and the restore drill.
-
Configuration reference
Every config section and key, the TOML → env → flag layering, and the
TELHA_*__-nested override rules. -
Key & secret management
API keys, token-signing keys, the field-encryption KEK, embedding/LLM credentials, and rotation flows.
-
Connectors
Entra ID, SharePoint/OneDrive, Exchange, Slack, Salesforce: app registrations, tokens, sync chains, ACL and deletion semantics.
-
Clarify bot
The Teams and Slack delivery surface:
CLARIFY_BOT_CONFIG, install, the delivery protocol, ask-in-chat, and web SSO. -
Backup & restore
Consistent snapshots of the RocksDB store and sidecar state, restore procedure, and verification.
-
Retention & archival
Retention policy, legal holds, archival of superseded versions, and the interaction with append-only history.
-
Day-2 operations
Key rotation, retention and legal holds, steward duties, LLM cost caps, and the eval release gate.
-
Troubleshooting
Symptom-to-cause tables: startup failures, DLL errors, auth, workers, generation, and vector search.
Read the runbook first
The Deployment runbook is written so an operator with no prior contact with the codebase can deploy and recover the system. Start there; the other pages go deeper on individual concerns.