Lenovo has published tech tip TT4152: XClarity Controller firmware built between 30 April 2025 and September 2026 can leave ThinkSystem V3 and ThinkAgile V3 servers with a dead BMC, no power-on, or a POST hang at "Waiting for BMC ready". Here are the affected builds, the fix versions per model and what to do before it bites.
Sysadmins started comparing notes late on Friday 16 October as Lenovo tech tip TT4152 circulated, with one operator describing it hitting many of the V3 servers deployed over the past two years and planning a day of hardware engineer call-outs. The fault sits in the XClarity Controller (XCC), the baseboard management controller on almost every ThinkSystem V3, ThinkAgile V3 and recent ThinkEdge server. This page sets out which models and XCC builds are affected, the fixed firmware for each platform, and the precaution Lenovo asks for on servers that have not yet failed.
What operators are seeing
Lenovo lists three symptoms, and a server can show one or several of them. Remote management stops working, so the XCC web interface, Redfish, IPMI and remote console all fail to connect. The server refuses to power on and the power LED flashes quickly, even after the power cords have been pulled and refitted. Or the server does start, but the fans run at full speed and the POST screen sits on "Waiting for BMC ready...".
The second and third symptoms are the serious ones, because they turn a management problem into downtime. A server that is running normally keeps running, but the next reboot, patch cycle or power event is when the BMC fault can stop it coming back. That is why the discussion this week is less about the bug and more about how quickly whole fleets can be updated.
Which servers are affected
The tip covers ThinkSystem V3 servers and ThinkAgile servers built on V3 platforms running XCC firmware built between 30 April 2025 and September 2026. That spans the mainstream Intel racks (SR630 V3, SR650 V3, SR850 V3, SR860 V3, SR950 V3), the AMD EPYC racks (SR635 V3, SR645 V3, SR655 V3, SR665 V3), the entry SR250 V3, ST50 V3 and ST250 V3, the ST650 V3 tower, the SD530 V3 to SD665-N V3 dense nodes, the GPU platforms (SR675 V3, SR680a V3 including the B200 version, SR685a V3, SR780a V3) and the ThinkEdge SE100, SE350 V2, SE360 V2 and SE455 V3.
On the HCI side, Lenovo names the ThinkAgile HX630 V3 to HX665 V3 Nutanix nodes, the MX455 V3 to MX650 V3 Azure Local systems and the VX630 V3 to VX850 V3 VMware nodes, including the DPU and SAP HANA variants. The SR258 V3, ST258 V3, ST58 V3 and ST658 V3 are on the affected list too, but the tip gives no version table for them.
ThinkSystem V4 servers use XCC3, not XCC2, and are not in the tip. V2 and older rack servers (SR630, SR650 V2 and so on) are not listed either. The ThinkEdge SE350 V2 and SE360 V2 are the exceptions, because they run XCC2.
Affected builds and fix versions: rack and tower servers
XCC firmware names start with a platform prefix (ESX for the Intel SR630 V3 and SR650 V3, KAX for the AMD 1U and 2U models, CTX for the entry servers). Check the version shown on the XCC firmware page against the range below. Lenovo pairs each XCC fix with a UEFI level, and its general guidance is to update XCC before UEFI.
| Server | Machine type | Affected XCC builds | XCC fix | Matching UEFI |
|---|---|---|---|---|
| SR630 V3 | 7D72, 7D73 | ESX350L-6.40 to ESX362K-8.10 | ESX362W-8.20 | ESE140G-3.91 |
| SR650 V3 | 7D75, 7D76 | ESX350L-6.40 to ESX362K-8.10 | ESX362W-8.20 | ESE140G-3.91 |
| SR635 V3 | 7D9G, 7D9H | KAX360M-5.40 to KAX374R-6.90 | KAX374T-6.91 | KAE146H-5.90 |
| SR645 V3 | 7D9C, 7D9D | KAX360M-5.40 to KAX374R-6.90 | KAX374T-6.91 | KAE146H-5.90 |
| SR655 V3 | 7D9E, 7D9F | KAX360M-5.40 to KAX374R-6.90 | KAX374T-6.91 | KAE146H-5.90 |
| SR665 V3 | 7D9A, 7D9B | KAX360M-5.40 to KAX374R-6.90 | KAX374T-6.91 | KAE146H-5.90 |
| SR850 V3 | - | RSX322M-7.50 to RSX332U-8.11 | RSX332W-8.20 | RSE128J-8.51 |
| SR860 V3 | - | RSX322M-7.50 to RSX332U-8.11 | RSX332W-8.20 | RSE128J-8.51 |
| SR950 V3 | - | EBX318K-6.50 to EBX328K-7.10 | EBX328V-7.20 | EBE124H-7.51 |
| SR250 V3 | 7DCL | CTX334G-6.20 to CTX344N-6.51 | CTX344P-6.60 | CTE128H-8.41 |
| ST50 V3 | - | CTX334G-6.20 to CTX344N-6.51 | CTX344P-6.60 | CTE128H-8.41 |
| ST250 V3 | - | CTX334G-6.20 to CTX344N-6.51 | CTX344P-6.60 | CTE128H-8.41 |
| ST650 V3 | 7D7A, 7D7B, 7D7C | USX3A4G-10.10 to USX3D6M-13.11 | USX3D6O-13.20 | USE144H-8.51 |
Dense, GPU and edge platforms
| Server | Machine type | Affected XCC builds | XCC fix | Matching UEFI |
|---|---|---|---|---|
| SD530 V3 | - | USX392J-5.10 to USX3D2L-8.11 | USX3D2N-8.20 | FNE130J-4.51 |
| SD535 V3 | - | QGX390J-8.10 to QGX3C2O-15.50 | QGX3C2Q-15.51 | GPE126I-5.51 |
| SD550 V3 | - | USX392J-5.10 to USX3D2L-8.11 | USX3D2N-8.20 | FNE130J-4.51 |
| SD650 V3 | 7D7M | USX3A4G-10.10 to USX3D6M-13.11 | USX3D6O-13.20 | USE144H-8.51 |
| SD650-I V3 | 7D7L | USX3A4G-10.10 to USX3D6M-13.11 | USX3D6O-13.20 | USE144H-8.51 |
| SD650-N V3 | - | USX3A4G-10.10 to USX3D6M-13.11 | USX3D6O-13.20 | USE144H-8.51 |
| SD665 V3 | 7D9P | QGX348L-9.11 to QGX3C2O-15.50 | QGX3C2Q-15.51 | QGE146K-8.52 |
| SD665-N V3 | - | QGX348L-9.11 to QGX3C2O-15.50 | QGX3C2Q-15.51 | QGE146K-8.52 |
| SR675 V3 | 7D9Q, 7D9R | QGX348L-9.11 to QGX3C2O-15.50 | TBD | TBD |
| SR680a V3 | - | R3X318M-6.10 to R3X334C-11.20 | TBD | TBD |
| SR685a V3 | - | R5X308R-3.10 to R5X320C-7.10 | TBD | TBD |
| SR780a V3 | - | R3X318M-6.10 to R3X334C-11.20 | TBD | TBD |
| SE100 | - | DZX314L-2.10 to DZX322L-4.11 | DZX322N-4.20 | DZE110F-1.41 |
| SE350 V2 | - | IYX342K-7.10 to IYX354N-9.21 | IYX354P-9.30 | IYE124F-6.50 |
| SE360 V2 | - | IYX342K-7.10 to IYX354N-9.21 | IYX354P-9.30 | IYE124F-6.50 |
| SE455 V3 | - | MBX324L-7.10 to MBX342L-9.21 | MBX342N-9.30 | MBE124E-6.51 |
For the SR675 V3, SR680a V3, SR685a V3 and SR780a V3 GPU servers the fix is still listed as to be confirmed. Those are the most expensive boxes on the list, so the precaution below matters most for them.
What Lenovo says to do
- If a listed server already shows one of the symptoms, Lenovo's instruction is to contact Lenovo Support. Do not keep power-cycling it; the tip says the fast-flashing power LED persists after an AC cycle.
- If a listed server is healthy, update XCC to the fix version for that platform, then apply the matching UEFI level.
- Until the update is applied, Lenovo recommends restarting the BMC twice in a row straight away as a precaution.
- Where the fix is still marked to be confirmed (SR675 V3, SR680a V3, SR685a V3, SR780a V3), the double BMC restart is the only published step for now. Check TT4152 again before any planned host reboot.
- The XCC firmware page shows both the primary and backup BMC versions. Check both, so a later failover does not bring back an affected build.
Why this matters for buyers, not just admins
The affected build window runs from April 2025 to September 2026, so a server that arrived with recent firmware and was updated as normal is no safer than one left alone. That includes new systems still in boxes, systems in a spares pool, and any second-hand V3 hardware coming onto the market from refreshes. Firmware level now belongs on the checklist for any V3 server that changes hands, next to warranty status and the XCC2 Standard or Platinum licence level.
The ETON view
For existing Lenovo estates this is a firmware job, not a hardware one, and it is fixable. The priority order is simple: healthy servers that are due a reboot soon go first, GPU servers get the double BMC restart now and the fix as soon as Lenovo publishes it, and anything already showing the symptoms goes to Lenovo Support rather than through repeated power cycles. If you run a mixed estate, put the XCC version check into the same patch window as this month's operating system updates so there is only one round of reboots.
For buyers, the practical change is that firmware level is now part of what you are paying for. Before you accept a ThinkSystem V3 or ThinkAgile V3 server from any supplier, us included, ask for the XCC primary and backup versions and the UEFI level in writing, and check them against the table above. That applies equally to new stock that has sat in a warehouse, to spares, and to refurbished V3 systems from refresh programmes. A server that ships on a fixed build costs nothing extra, and one that ships on an affected build can cost you an engineer visit.
We stay vendor-neutral on this. HPE iLO 6 and Dell iDRAC9 have their own advisory streams, and the same rule applies: know the BMC version on every server you own and every server you buy. If you are weighing whether to keep standardising on ThinkSystem V3 or move a refresh to HPE ProLiant Gen11 or Dell PowerEdge 16G, send us the model list and we will price the like-for-like alternatives alongside Lenovo, with current UK availability and lead time, and quote buyback on any hardware you retire.
Related infrastructure
- Lenovo ThinkSystem Rack Servers
- Lenovo ThinkAgile HCI
- Lenovo ThinkSystem AI & GPU Servers
- Lenovo Dense and Multi-Node Servers
- Lenovo Edge Servers
Category: Servers · Vendor: Lenovo · Technology: XClarity Controller 2 (XCC2), ThinkSystem V3, ThinkAgile V3, BMC firmware · Last verified: 17 Oct 2026 · ~7 min read
Sourcing this kind of infrastructure? Talk to ETON about availability, lead time and pricing.
