Problem

Redis starts with zero configuration — no redis.conf, no command-line arguments. Default maxmemory is 0 (unlimited). Redis is used as Celery broker, results backend, and filter state cache. A spike in SQL Lab queries can consume all available memory, OOM-killing Redis and any co-located containers.

File: src/container.py (Redis class, lines 376–390)

Impact

Redis OOM → Celery broker dies → all async queries fail → Superset appears frozen.

Fix

Create services/superset/redis.conf:

maxmemory 512mb
maxmemory-policy allkeys-lru
tcp-backlog 511
timeout 300

Mount into Redis container. Also apply Redis admin guide recommendations in the entrypoint:

echo never > /sys/kernel/mm/transparent_hugepage/enabled
sysctl vm.overcommit_memory=1

Acceptance Criteria

• services/superset/redis.conf created with maxmemory=512mb, maxmemory-policy=allkeys-lru, timeout=300
• Redis container mounts redis.conf and starts with redis-server /etc/redis/redis.conf
• Transparent huge pages disabled and vm.overcommit_memory=1 set in container entrypoint