RAID Controller Battery Failed – Can You Still Boot?

Quick Answer

Yes, the server will boot. The RAID controller just disables write caching to avoid data loss if power fails. Performance will drop hard—expect 50-80% slower writes.

Context: Why This Happens

Here's what's actually happening. RAID controllers use a small battery (or supercapacitor) to power the write cache in RAM. If the battery dies or fails its learn cycle, the controller assumes it can't protect that cached data during a power loss. So it flips write caching off. You get a warning at POST like Battery backup unit failed – write cache disabled or BBU status: Failed.

This is not a problem for reads. Your system will still boot and serve files. But anything that writes to the RAID—databases, log files, file copies—will feel sluggish. On a Dell PERC H710P I saw write throughput drop from 800 MB/s to 120 MB/s after a battery died.

The battery itself is a consumable part. They have a lifespan of 2-5 years depending on temperature and how often the controller does a learn cycle (recharge/discharge to calibrate capacity). A dead battery doesn't mean the RAID is damaged. It just means the controller is being safe.

Fix Steps

Check the battery status – Boot into the RAID BIOS (usually Ctrl+R on Dell, Ctrl+H on HP/LSI). Look for the battery section. You'll see something like BBU Status: Failed or Learn cycle failed. This confirms the battery is the problem, not the controller.
Order a replacement battery – Get the exact part number for your controller. Dell PERC H710 uses part number 0C7T0F. HP Smart Array P420 uses 649690-001. Do not guess. A wrong battery can cause charging issues or damage the board.
Replace the battery – Power down the server, unplug power cords, open the case. The battery is usually a small rectangular pack on a cable, plugged into the RAID card. Disconnect the old one, snap in the new one. No soldering. No special tools. Takes 2 minutes.
Let it charge – Boot the server. The controller will start recharging the new battery. This takes 4-12 hours depending on the capacity. You can use the server during this time, but write cache stays off until the battery reaches full charge and passes a learn cycle.
Run a learn cycle – After charging, the controller auto-runs a learn cycle (discharge/recharge). This calibrates the battery's capacity. It takes another 4-8 hours. During this time, write cache will still be off. Once the learn cycle passes, write cache re-enables automatically.

Alternative Fixes If the Main One Fails

Option A: Force Write Cache On Without Battery

If you absolutely can't replace the battery right now and performance is critical, you can override the safety check. But this is risky. If power drops, anything in the write cache is gone. RAID metadata can corrupt and the whole array might fail to mount.

On Dell PERC: boot into CTRL+R, go to Virtual Disk Management, select the virtual disk, press F2, choose Cache Policy, set it to Write Back and confirm the warning. On HP Smart Array: use the ACU (Array Configuration Utility) in the OS, select the logical drive, change write cache policy from Write Through to Write Back.

Only do this if the server is on a UPS and you can tolerate a possible array repair. I did this once on a backup server that had no battery option. It worked for two years, but when power died during a learn cycle, the array lost the last 30 seconds of writes. We had to rebuild from tape.

Option B: Replace the Entire RAID Card

If the battery is soldered in or the card is old (e.g., PERC 5/i, 6/i), a new battery might cost more than a used card on eBay. A used PERC H710 can be had for $30-50. Swap the card, import the foreign config (the RAID metadata on the disks), and you're back with full performance. The controller reads the existing RAID settings from the drives themselves. You don't lose data.

Option C: Disable BBU Monitoring

Some controllers let you disable the battery check entirely. On LSI 9260/9265 series, the MegaRAID Storage Manager has an option Disable BBU learn cycle. This stops the controller from complaining, but it also means you get no warning if the battery actually fails. The write cache stays on regardless. This is a band-aid, not a fix.

Prevention: Make the Battery Last Longer

RAID controller batteries hate heat. Keep server room temp below 25°C (77°F). Also, check the learn cycle schedule. Some controllers run a learn cycle every 30 days by default. That's fine. But if the battery never completes a learn cycle (e.g., server rebooted mid-cycle), it can degrade faster. So let the learn cycle finish. Don't force a reboot or shutdown during it.

Replace the battery proactively every 3 years if the server runs 24/7. Mark it in your calendar. A $40 battery is cheaper than two hours of performance loss during a critical backup window.

If you're building a new server, skip RAID cards with batteries. Use NVMe or SATA SSDs with on-board power loss protection (like Samsung PM9A3 or Intel D5-P4320). They don't need a separate battery. The capacitor on the SSD itself handles flush on power loss.

Takeaway: A dead RAID battery is annoying, not a disaster. Replace it, or live with slow writes. Don't force write-back unless you know the risks. Your data is worth more than a few seconds of speed.