Why Redis Exits on ARM64: Fix the THP COW Bug and Prevent Data Loss

Are you seeing this? Running Redis on Huawei Cloud, Alibaba Cloud ARM instances, Raspberry Pi, or Apple Silicon and the startup suddenly aborts with a warning.

Redis will now exit to prevent data corruption.
Note: you can suppress this warning by setting:
ignore-warnings ARM64-COW-BUG

You want a quick fix but worry: will skipping the warning cause data loss?

✅ Is this bug dangerous?

✅ When is it safe to ignore?

✅ How to correctly configure ignore-warnings ARM64-COW-BUG?

✅ How to adjust the container environment?

🔍 Core Issue: Redis "actively brakes" instead of crashing

The error originates from a known compatibility problem between Redis on ARM64 and Linux Transparent Huge Pages (THP):

When Redis performs an RDB snapshot (bgsave),

If the system has THP enabled,

On ARM64 this may trigger a kernel COW (copy‑on‑write) exception,

Resulting in corrupted RDB files or even memory data chaos.

Since Redis 6.0+, on ARM64 it forcibly checks THP status at startup; if a risk is detected, it exits (not just warns) to prevent silent data loss.

🛠️ Solution: Two Paths, Choose as Needed

✅ Path One: Recommended – Disable Transparent Huge Pages (Fix the Root Cause)

This is the official Redis recommendation and eliminates the risk completely.

# Temporary disable (effective immediately)
echo never | sudo tee /sys/kernel/mm/transparent_hugepage/enabled
echo never | sudo tee /sys/kernel/mm/transparent_hugepage/defrag

Advantage : One‑time fix, does not affect Redis functionality.

Verification : cat /sys/kernel/mm/transparent_hugepage/enabled should show [never].

Permanent disable : See appendix for systemd/rc.local method.

⚠️ Path Two: Cautious – Ignore the Warning (Work‑around)

Only consider this when all of the following are true:

Add the following line to redis.conf:

# Resolve ARM64 kernel COW bug warning
ignore-warnings ARM64-COW-BUG

After editing, restart Redis to start normally.

Container Users Note

Copy the config from the container:

docker cp <redis_container>:/etc/redis/redis.conf ./

Edit the local redis.conf and add the above line.

Mount it back when starting the container:

docker run -v $(pwd)/redis.conf:/etc/redis/redis.conf redis redis-server /etc/redis/redis.conf

Important Reminder :

This setting does not fix the kernel bug; it only silences Redis.

If you still use RDB, the data‑corruption risk remains.

🧪 How to Verify Your Setup Is Safe

Check if RDB is used:

grep -v "^#" /etc/redis/redis.conf | grep "save"
# Empty output or only "save \"\"" → RDB disabled → ignoring warning may be OK
# Presence of "save 900 1" etc. → RDB enabled → strongly recommend disabling THP

Check THP status:

cat /sys/kernel/mm/transparent_hugepage/enabled
# Shows [never] → safe
# Shows [always] → high risk; even with ignore‑warnings you should not run RDB

📜 Appendix: One‑Click Detection Script (with Risk Tips)

#!/bin/bash
echo "🔍 Detecting Redis ARM64 safety status..."
THP=$(cat /sys/kernel/mm/transparent_hugepage/enabled)
RDB=$(grep -v "^#" /etc/redis/redis.conf 2>/dev/null | grep -c "save [0-9]")

echo "THP status: $THP"
if [[ $THP == *"[always]"* ]] && [ $RDB -gt 0 ]; then
  echo -e "
🔴 High risk! THP enabled + RDB active, do NOT ignore warning!"
  echo "👉 Please run: echo never > /sys/kernel/mm/transparent_hugepage/enabled"
elif [[ $THP == *"[always]"* ]] && [ $RDB -eq 0 ]; then
  echo -e "
🟡 Medium risk: THP enabled but RDB not used."
  echo "👉 You may add 'ignore-warnings ARM64-COW-BUG', but disabling THP is still advised."
else
  echo -e "
🟢 Safe: THP disabled, Redis can run normally."
fi

🌟 Final Thought

Redis would rather refuse to start than let you lose data.
That stubbornness is why it is trustworthy.

In the ARM64 era, understanding the underlying mechanism is far more important than blindly skipping warnings.