playbook-ttl-not-expiring.md

Playbook: TTL Not Expiring

Scope

Hot and warm memory tiers.

Explanation

TTL issues lead to stale data and unbounded memory growth. This playbook ensures TTL is applied consistently in hot tier and enabled in Cosmos.

Triggers

• Stale data served
• Memory growth with low eviction

Triage

1. Verify TTL is set on write.
2. Check Redis key TTL and Cosmos TTL policy.
3. Identify keys missing TTL.

Mitigation

• Fix TTL settings in write paths.
• Enable Cosmos container TTL if disabled.
• Run cleanup for stale keys.

Prevention

• Add unit tests for TTL behavior.
• Enforce TTL defaults at builder level.

Implementation Steps

1. Set TTL explicitly on all hot writes.
2. Enable TTL at the Cosmos container level.
3. Add tests verifying expiry.

Code Examples

Redis TTL write

async def set_session(key: str, value: str):
	await hot_memory.set(key, value, ttl=300)

TTL test

import asyncio

async def test_ttl_expiry():
	await hot_memory.set("k", "v", ttl=1)
	await asyncio.sleep(2)
	assert await hot_memory.get("k") is None

Mermaid: TTL Remediation Flow

flowchart TD
	A[TTL not expiring] --> B[Inspect write paths]
	B --> C{TTL set?}
	C -->|No| D[Fix hot writes]
	C -->|Yes| E[Check Cosmos TTL settings]
	E --> F[Enable container TTL]

Loading

Escalation

If data growth threatens capacity, scale tiers and notify platform owner.