Analyze Memory State

Analyze Redis Memory

Use this skill when Redis memory usage matters: high memory, evictions, or storage optimization.

Workflow

1. Get the overview

• info — read used_memory, maxmemory, evicted_keys, expired_keys.
• dbsize — total key count.
• Note whether eviction is actively happening (evicted_keys growing).

2. Profile by type

• Use the server's scan + type tooling to count keys per type (string, hash, list, set, zset, stream). A single type dominating usually points at the culprit pattern.

3. Find the biggest consumers

• Strings: sample long values (strlen via get when feasible) on the largest keys.
• Hashes/lists/sets/zsets: estimate by element count (hlen, llen, scard, zcard) — element count × element size is the practical proxy; flag keys with extreme counts.
• Streams: xlen for very long streams with no trimming.

4. Identify candidates

Rank by estimated bytes:

1. huge values (large blobs, caches of large payloads),
2. unbounded collections (lists/streams that never trim),
3. keys with no TTL that should expire (session/cache keys without expiry).

5. Report

Give: memory overview (used/max/evictions), the top N keys by estimated size with their types and element counts, and concrete recommendations (add TTL, trim streams, move blobs out of Redis). Do not delete, expire, or evict anything — those are write actions requiring explicit user intent.