If two or more of your Antminer S9i units keep dropping off the network, swap IP addresses, or only ever show up one-at-a-time in your scanner, the cause is almost always a duplicate MAC address. This is commonly caused by the control board’s Ethernet MAC not being uniquely assigned, so multiple units advertise the same hardware address on your LAN. Your switch and router can only route to one device per MAC, so the units fight over the same ARP entry and DHCP lease. The fix is to flash each affected miner with a firmware image that regenerates a unique MAC, done one unit at a time in network isolation so the collision doesn’t hide which board you’re actually talking to.
These instructions apply to the S9i specifically. The plain S9, S9j, and S9 SE store and derive their network identity slightly differently — don’t assume the same firmware image or behaviour carries across the whole S9 family.
Why a duplicate MAC breaks your hashcenter
Every network device is supposed to carry a globally unique 48-bit MAC address burned into its network interface. Switches build their forwarding table by MAC, and routers resolve IP-to-MAC via an ARP cache. When two S9i units present the identical MAC:
- The switch overwrites its forwarding entry every time the “other” miner transmits, so traffic ping-pongs and one unit effectively disappears.
- DHCP hands the same reservation to both, producing IP conflicts and web UIs that load the wrong miner.
- Your scanner lists a single device where two are physically wired in, or the reachable unit flips at random.
This is a network-identity problem, not a hashboard fault. The chips (the S9i runs BM1387 ASICs across its three hashboards) and the PSU are fine — only the control board’s address assignment is wrong.
Confirm it really is a duplicate MAC
Before flashing anything, verify the diagnosis so you don’t chase a cabling or DHCP-scope problem instead:
- Wire up two suspect miners and let both boot.
- From a PC on the same subnet, read the ARP table: run
arp -a(Windows/macOS) orarp -n(Linux). - If two different IPs resolve to the same MAC address — or if only one IP ever appears no matter how you power-cycle — you have a duplicate MAC.
- Cross-check in each miner’s web UI under System → Network / Overview: if both units display an identical MAC string, it is confirmed.
What you’ll need
- A router or managed switch running DHCP, plus a laptop/PC on the same subnet.
- Two or more Ethernet cables (work with one miner connected at a time regardless).
- A LAN ASIC scanner / bulk-management tool (Bitmain’s IP Reporter or any equivalent miner-scanner utility).
- The correct S9i firmware image. Pull it from Bitmain’s official support/download portal for the ANTMINER S9i, or from our manuals and downloads library. Do not flash a generic S9 image onto an S9i.
- Default web-UI credentials (Bitmain stock firmware ships
root/root) unless you’ve changed them.
The fix, step by step
Isolate first. Because the units collide on the same MAC, you cannot reliably work on them while both are on the LAN — you’ll never be sure which board you’re flashing. Physically disconnect all but one affected miner.
- Connect a single miner. One S9i, one Ethernet run to the router/switch, one PC on the same subnet. Wireless-to-the-router is fine; the miner itself is wired.
- Locate it. Run your scanner tool and find the miner’s current IP. (If it won’t pull an IP, press the IP Report button on the control board to force it to announce itself.)
- Log in. Open the miner’s IP in a browser and sign in to the web UI.
- Record the current MAC. Note it from System → Network — you’ll compare against this after flashing to prove the address changed.
- Flash the firmware. Go to System → Upgrade, select the correct S9i image, and apply it. Reflashing regenerates the network identity regardless of the “keep settings” toggle (which only governs whether your pool and config are retained). Do not cut power during the flash — an interrupted write can brick the control board and turn a five-minute network fix into a board-level repair.
- Reboot and re-scan. After it restarts, the miner requests a fresh DHCP lease under its new MAC, so its IP will change. Re-run the scanner and reconnect at the new address.
- Verify the new MAC. Confirm the address under System → Network is different from what you recorded in step 4.
- Repeat per miner. Power down, connect the next affected unit in isolation, and run the same sequence. Every unit must end up with a distinct MAC.
- Reconnect everything. Wire all miners back onto the LAN together.
How to confirm it worked
- Every affected S9i appears simultaneously in your scanner — no more units dropping in and out.
arp -anow shows a unique MAC for each IP, with no two entries sharing an address.- Your router stops logging IP-conflict warnings, and each web UI loads the miner you expect.
- Each miner holds a stable, distinct DHCP lease across reboots.
Common mistakes
- Flashing both miners on the same LAN at once. The collision means you can’t tell which board you’re addressing — isolate, always.
- “Fixing” it with static IPs. Assigning fixed addresses doesn’t help: the switch still forwards by MAC, so identical MACs keep colliding at layer 2. You must change the MAC, not just the IP.
- Wrong image. A plain-S9 firmware on an S9i can leave you worse off. Match the image to the model.
- Skipping the “before” MAC. Without recording the original address you can’t prove the flash actually changed anything.
When to escalate
If a clean flash completes but the MAC is unchanged, or the unit won’t hold a fresh address after several attempts, the control board’s stored identity may be corrupt or the board itself may be failing — that’s past a firmware fix. Run the miner through our ASIC fault finder to rule out a deeper control-board fault, and if it points to hardware you can start a repair with us. Related: our full manuals library covers S9-series firmware and setup end to end.
