On a used Broadcom 9361-8i, the most common silent failure is not the controller itself — it is the CacheVault supercapacitor. Check it with storcli /c0/cv show. If the state reports Degraded or Failed, the controller drops write-back cache and array write performance falls with it. The fix is a replaceable module, not a new 9361-8i RAID controller.
What the CacheVault actually does
The 9361-8i is a 12Gb/s SAS/SATA controller on PCIe 3.0, shipped with either 1GB or 2GB of cache — for example the 9361-8i 1GB cache and 9361-8i 2GB cache variants. Older LSI generations protected that cache with a lithium-ion battery pack (the iBBU), which was the part most likely to swell and die. The 9361-8i generation replaced the battery with CacheVault: a supercapacitor plus NAND flash. During a power loss, the supercap keeps the controller alive just long enough to flush the cache to flash, and the contents are written back to the array at the next boot.
There is no battery chemistry involved, nothing swells, and service life is much longer than an iBBU. But a supercapacitor is still a wear part: its capacitance fades with heat and age, and it eventually can no longer hold the controller up long enough to complete a flush.
| Cache protection | Power source | Common failure on used units | What happens when it fails |
|---|---|---|---|
| iBBU (older LSI cards) | Li-ion battery pack | Capacity loss, swelling | Write-back cache disabled |
| CacheVault (9361-8i) | Supercapacitor + NAND flash | Capacitance fades with heat and age | State turns Degraded; write-back drops to write-through |
| No module installed | None | — | Write-through only; lower write performance |
The states you will see, and what each one means
Optimal is what you want: full capacity, cache fully protected. Degraded means capacitance has fallen below the threshold the controller accepts for a safe flush — the card typically switches from write-back to write-through, so the array keeps running but writes get slower. Failed means the supercap can no longer support a flush at all; a power cut at the wrong moment can lose the data sitting in cache.
Do not confuse a dead supercap with a dead controller. The main board and the cache/flash module are usually fine — the supercap is a separate, replaceable part bolted onto the cache module. On a degraded card, replacing the kit restores full function.
How to check it with storcli
If the card is already installed in a system, run:
storcli /c0 show — the controller overview includes the CacheVault status line.
storcli /c0/cv show — CacheVault detail: state, temperature, and remaining capacitance.
Read the State field first. Anything other than Optimal deserves a closer look at remaining capacity and temperature. You will also see it at boot: many systems print a “CacheVault mode degraded” warning during POST, which is easy to miss on a headless rack. If you are buying rather than diagnosing, ask the seller to run these two commands before the card ships to you — a seller who tests on an in-house bench can quote you the CV state as part of the test result.
Why this matters more when the card is used
A used 9361-8i may have spent years in a hot chassis, and supercapacitors age with heat. Two cards of the same model and the same cache size can have very different remaining supercap life, so “9361-8i” alone does not tell you the condition — the CV state does. When comparing offers, the difference between a card with an Optimal CacheVault and one reporting Degraded should be roughly the cost of a replacement kit, because that is the routine fix.
One boundary worth stating plainly: if your workload depends on protected write cache from day one and you cannot test the card before deploying it, the CV state is not a detail — it is the item to verify first. If the card is for a read-heavy or write-through-tolerant role, a worn supercap is a smaller deal and can be fixed on your schedule.
If you are still choosing between generations, our comparison of MegaRAID 9361 vs 9460 vs 9560 covers where the 9361-8i sits, and 9460-8i 2GB vs 4GB cache covers the cache-size question on the newer Tri-mode family.
Replacing the module
Match the kit to the controller family. The 9361-8i is SAS3108-based and takes the CVM02-style CacheVault kit, such as the CVM02-8G supercap kit and the CVM02 4GB kit. Note that the 1GB/2GB cache size lives on the cache module itself, not the supercap kit — replacing the supercap does not change your cache size. Module part numbers are not interchangeable across controller families, so verify before you order.
FAQ
How long does a CacheVault supercapacitor last?
It depends on operating temperature and powered hours, not on a fixed date. Supercaps outlast the old iBBU batteries by a wide margin, but capacitance fades steadily. Read the CV state instead of guessing from the card’s age.
Can I run a 9361-8i with a dead CacheVault?
Yes. The controller runs its cache in write-through mode, so the array keeps working, but write performance drops and cache contents are not protected through a power cut. Replacing the module restores write-back.
Does a Degraded CacheVault mean the controller is bad?
No. The supercapacitor is a separate replaceable part. If the controller passes diagnostics and the cache/flash module is intact, a new kit restores full function.
How do I know which CacheVault kit fits my card?
Match the kit to the controller family — the SAS3108-based 9361-8i uses the CVM02-style module. Send the part number printed on your cache module and we will confirm the match.
Is this the same on the 9460-8i and 9560-16i?
The principle is the same on Tri-mode cards — supercap plus flash — but the module part numbers differ by family. Always verify by part number before ordering a replacement.
Buying a used 9361-8i or a CacheVault kit? Send the part number and quantity and we will check availability. WhatsApp +86 13068616914 · zhongxindata.hk
What the CacheVault warning actually means
- The controller charges its supercap continuously and uses it to hold the write cache through a power event
- A warning usually means the capacitor has lost capacity, so the hold-up time is no longer sufficient
- The card keeps working, but it drops out of write-back mode, which changes performance and alerting
Testing the supercap before you deploy
- Read the controller’s own supercap status through the management interface
- A controlled power-cycle test confirms the controller survives on battery as it should
- Replace rather than ignore — a failing supercap is a known, cheap part, not a controller fault
What to do with a used controller that shows the warning
- Treat the warning as a condition disclosure: we report it per unit rather than hiding it
- A card with a healthy or replaced supercap is a straightforward purchase
- Confirm firmware level as well, since controller behaviour differs between releases
Related reading
- MegaRAID 9361-8i — the card this article is about
- 9460 cache comparison — the related cache question
- Data center RAID strategy — broader controller guidance
- RAID controllers — category hub
Frequently Asked Questions
What does the CacheVault warning on a 9361-8i mean?
It usually means the supercap has lost capacity. The controller charges the capacitor continuously so it can preserve the write cache during a power event, and a warning indicates the remaining hold-up time is no longer sufficient. The card still works, but it runs in write-through mode instead of write-back, so write performance drops and the controller alerts differently.
Is a 9361-8i with a CacheVault warning unusable?
No. The warning is about the supercap, not the controller logic. The supercap is a small, widely available part, so a replacement restores full write-back protection. We report the supercap status on every unit we list so you know the state before you buy rather than discovering it at install time.
How do I test the supercap myself?
Read the supercap status from the controller management interface first, then do a controlled power-cycle test: the controller should survive the power loss and come back with the cache intact. If it does not, the supercap is not holding charge and should be replaced.
What happens to performance with a failing supercap?
The controller falls back to write-through, so every write waits for the media. On a spindle array that is a visible throughput drop; on flash it is less severe. The bigger practical effect is that you lose the protection that lets the controller acknowledge writes early, which is the reason the supercap exists.
How do you verify a used 9361-8i before shipping it?
We clean, re-flash and bench-test on our own test line, then confirm the part number, cache size, connector count, port mapping and firmware level, and we read the CacheVault supercap status. If a unit has a supercap issue we say so, and we can quote it as-is or with the supercap addressed. Ask for photos of the actual card before you commit.
Does the firmware version affect CacheVault behaviour?
Controller behaviour does differ between firmware releases, and we state the level each used unit ships with so you are not inheriting an old image by accident. We re-flash during testing, so the level is one we have verified on our own test line rather than whatever the previous owner left behind.