Latest questions from the community.
An HPE FlexFabric 5950 48SFP28 8QSFP28 (JH402A) always downgrades an HPE JH419A 100G QSFP28 LC SWDM4 transceiver to 40G and never brings it up at full speed. Why?
A 1000BASE-T copper SFP (Arista SFP-1G-T) in an Arista switch does not link to a device that only rnus 100 Mbit/s, and a 10GBASE-T SFP+ does not fall back to a lower rate either. How does EOS handle speed negotiation for copper SFP modules?
An M290 with the optional two-port fibre module was fitted; can ports A0 and A1 be used as ordinary interfaces, for example for an ISP fibre handoff, and can a 1 Gb SFP be used in them?
What do the port states in the switchshow output of a Brocade Fibre Channel director or switch mean, and which of them point at the transceiver rather than the fabric?
A UCS 6332 fabric interconnect has only 40G QSFP+ ports, but the upstream switches are 10G. How can the FI be uplinked at 10G, and why does the breakout option not appear on some ports?
With auto-negotiation off and 10G configured on sfp1/sfp2 of a CCR1036-12G-4S (RouterOS 6.43.8), MikroTik S+DA0001 and Amphenol DACs still run at 1G and the Cisco side shows no link at 10G. Which part of this link should be checked first?
Cisco troubleshooting document: inserting certain SFPs (for example GLC-SX-MMD or third-party modules) into a Catalyst 3850 C3850-NM sub-module produces %PLATFORMPM-6-MODULEERRDISABLE: The inserted SFP module ... is not supported and %PM-4-ERRDISABLE: gbic-invalid ... putting ... in err-disable…
The poster asked how to test fibre cables from a switch. The guidance covers reading optical levels with show interfaces transceiver / show controllers, comparing TX and RX at both ends, and recognising that a reading of -40, -50 or -80 dBm means no signal is being received. Hardware involved…
A D-Link DES-3200-28 B1 switch logged DDM Port X optic module RX power exceeded the low warning/alarm threshold; the optical level looked abnormally low in the switch although the module measured fine when tested standalone. Hardware involved: D-Link DES-3200-28 B1, DDM-capable SFP. Reported…
Two Jetson Thor devices achieved 100G connectivity directly but with poor speed, and when each was connected to a 100G switch no link formed, although the same optics and cables passed a 100G loopback test against a different switch. The poster wondered whether insufficient speed was breaking…
The poster could not find documentation for the command service unsupported-transceiver and asked what it does and whether there are consequences. It is one of Cisco's hidden commands: without it IOS disables a port when a non-Cisco transceiver is inserted, because the OS reads the vendor field and…
A Huawei S6700 iStack did not form because the optical modules on the dedicated stacking ports 45-48 were of different types; the platform requires identical transceivers on those ports. The four ports 45-48 are the fixed stacking ports, and mixing, for example, an SR module on one and an LR module…
FS Box V2 is a firmware/coding tool kit for optical transceivers and DAC/AOC cables. It reprograms a module's compatibility code so it is accepted by a target switch platform, letting engineers adjust coding in the field or when a switch is upgraded to a version that no longer accepts the old code…
A customer with a 10Gtek ASF-10G-T 10GBASE-T transceiver could not read its EEPROM: ethtool -m ens802f0 returned Cannot get module EEPROM information: Operation not supported. The vendor attributed it to a firmware version mismatch between the NIC and the transceiver. Hardware involved: 10G NIC…
A proposed ixgbe patch would have let X550EM NICs recognise otherwise-unsupported 1G copper SFPs by treating them as 1G SX through a new module parameter, mirroring the existing unsupported-SFP handling. Without the change these NICs simply refuse the copper module, so operators wanting cheap 1G…
Huawei switches with third-party optics log SRM/3/SFPEXCEPTION ... Optical module exception, SFP is not certified daily, cluttering the logs during migrations even though the modules work. The modules pass traffic normlaly, but the daily SFPEXCEPTION traps flood the log buffer and the syslog…
On Huawei switches with 25GE ports the poster asked about using SFP+ transceivers and copper cables; per the troubleshooting docs, when a non-1 m copper cable is connected to a 25GE interface the RS-FEC function must be enabled or the interface goes Down, and mixing different optical modulles at…
Some Cotsworks SFF fibre modules ship with invalid data in the first EEPROM bytes (physical ID, extended physical ID, connector type) and with checksums programmed before the fields were finalised, so the Linux phy/sfp driveer does not detect them as valid modules. A kernel fixup rewrites the three…
An ODI DFP-34X-2C2 GPON ONU SFP failed to advertise its 2500BASE-X capability because its EEPROM (or virtual EEPROM) did not provide correct speed information to the Linux sfp driver, so the interface did not run at the expected rate. A kernel patch adds a quirk to force the right speed. Hardware…
On a Mellanox MSN4600C (ACS-MSN4600C) running SONiC in a T1 topology, the test testchcksfpusingethtool failed because ethtool -m sfp[ID] returned rc=1 with Cannot get module EEPROM information: No such device, i.e. the module EEPROM was not accessible on that platform during the test. Hardware…
On a Banana Pi BPI-R4 running OpenWRT 24.10.0 an SFP module failed with kernel logs failed to read EEPROM: -ENXIO and warnings that the module was slow to respond. The internet still worked but the log filled with the error, and the poster was unsure whether it pointed to a faulty stick, a GPON…
On an Accton Wedge100BF-32x the ONLP call onlpsfpeepromread returned QSFP EEPROM data with alternating bytes skipped, so vendor strings appeared with every other symbol missing; the same transceiver read correctly on an Accton AS7312-54x. The vendor name FINISAR CORP appeared at the wrong offset…
Adapter boards that, with a serial programmer, read and write transceiver memory across SFP, SFP+, SFP28, SFP112, QSFP/QSFP+/QSFP28, SFP-DD, QSFP-DD, XFP and OSFP. The software covers the MSA/CMIS memory maps, manages host/manufacturer passwords including brute-force search, changes…
An open hardware/firmware project to read and write the EEPROM inside SFP and SFP+ modules so their capailities and vendor origin can be inspected and modified. It uses an RP2040 microcontroller and exposes a USB Mass Storage interface. The project is a work in progress and does not yet document…
The poster wanted to flash four TRW-1310-1.25C-3 and TRW-1550-1.25C-3 SFP modules to work with an HP v1910-24G (JE006A) and asked about a programmer. The difficulty was finding the right firmware and matching serial numbers within one switch while keeping checksums valid. Hardware involved: HP…
A Python library to create, read and modify the 256-byte A0h page of SFP EEPROM per SFF-8472. It exposes named fields (vendor, part, serial, wavelength, bit rate, encoding, link lengths, diagnostics options), generates images from scratch or from a binary, and validates/recalculates checkssums. It…
A user guide for an SNR SFP/SFP+ reprogramming solution: USB programmer plus Windows/Linux software (FTDI drivers required) supporting SFP and SFP+ per INF-8053, SFF-8472/8431, INF-8077i and INF-8438. The workflow is connect, read the module to display vendor/model/wavelength, load a new firmware…
A USB coding-board product line to reprogram optical transceivers so they match different equipment vendors. Supported form factors span GBIC, SFP, SFP+, SFP28, XFP, QSFP, QSFP28, QSFP-DD plus RJ45, DAC and AOC. The software reads, decodes, edits and writes module memory, ships about 400 well-known…
A web-based (Web Bluetooth) GUI for the UniFi SFP-Wizard device that reads SFP EEPROM, saves backups as .uieeprom files and writes from a prior backup. Limitations: the API does not flash directly; the dump goes to the device snapshot buffer for manual confirmation on the hardware; newer firmware…
A set of bash scripts using i2c-tools to modify SFP module EEPROM. Hardware is an STM32 configured as a USB-I2C adapter (or a Raspberry Pi / mainboard SMBus) with a Cisco Twingig adapter as the SFP breakout. It handles the default Cisco Finisar module password 0x00000000 and recalculates checksums…
A Raspberry Pi tool for SFP/SFP+ EEPROM: -r prints connector, transceiver spec, wavelength, vendor, part, serial, date and checksums; -d dumps the whole EEPROM; -c recaculates checksums and offers to write the correct value; -m reads DOM sensors (temperature, voltage, bias, power). A vendor variant…
An open-source hardware and software project to read and modify SFP/SFP+ EEPROM, aimed at changing the OUI and other identity fields so a module works with third-party transceivers where the vendor has not fully disabled write access. It uses a Cypress CY8CKIT-059 board as the USB interface. As of…
A tutorial on recoding SFP/SFP+ EEPROM so a module works with a different vendor's switch. It uses a Cisco Twingig housing to reach the SFP I2C pins and wires four lines from the Pi GPIO (3.3 V, GND, SDA on GPIO3, SCL on GPIO5). It shows detecting 0x50 (and optional 0x51 DOM) with i2cdetect, points…
Discussion of why re-coded modules fail Cisco validation: Cisco SFPs carry a unique signature tied to each serial number in bytes 96-127, and editing vendor name or serial makes the signature no longer match, so the switch rejects the module unless the signature is regenerated. Hardware involved…
A small toolset that reads and writes SFP EEPROM over I2C with a CH341A programmer plus an I2C sniffer. The author found that sigrok/PulseView did not work with the analyser, that a v1.0.5 SFP coder firmware reported success while silently not writing the module (proven by I2C traces), and that the…
A reference post on the memory layout used when programming transceivers: SFP blocks A0h (serial ID/vendor, often write-protected) and A2h (diagnostics, page-select byte 7Fh, page 00h/01h user-writable); QSFP block A0h with page-select 7Fh, page 00h identity write-protected and page 03h user…
Reverse-engineering notes on the SFP Wizard: it reads a module's part number (page A0h bytes 40-55), looks it up in an embedded password database and sends the matching unlock password, falling back to brute force when unmatched. It traces the firmware evolution: v1.0.5 tried only the first…
A notes repository on hacking SFP/QSFP EEPROM: SFPs present EEPROM at I2C 0x50 and DOM plus vendor/password features at 0x51; locked moddules need a 4-byte password written to 0x51 bytes 0x7b-0x7e. It lists several known unlock passwords and evaluates write methods (off-the-shelf coder limited to…
The poster programmed an OEM copper SFP with a 256-byte firmware dump; the switch recognised the type but the module received no data, whereas the original worked, and no password error appeared. It failed on several units. Hardware involved: OEM copper SFP (24C02 EEPROM). Reported…
Users needed to write to password-protected SFP/QSFP/XFP modules from HP, Cisco, Finisar and JDSU. The thread documents the two protection levels: a host/user password protecting the user-writable area (A2h offset 7Bh on SFP, A0h offset 7Bh on QSFP/XFP) and a manufacturer password protecting the…
After a power outage an Aruba Instant On 1930 (24 ports plus 4 SFP) still recognised inserted SFP modules but every SFP port stayed down; inserting a new transceiver made the port LED blink once and then nothing. The poster asked for diagnostics or replacement options. Hardware involved: Aruba…
A Meraki MX95's WAN2 SFP would not link to an upstream Extreme switch, while the same SFP and patch linked to a Netgear switch, which in turn linked to the Extreme. The same happened with an MX85 against an AT&T-provided Ciena device. Cisco-certified and Extreme-certified SFPs were tried in all…
A pre-built HL15 server (Supermicro X11SPH-nCTPF wth Intel X722 SFP+ ports, Rocky Linux) would not link to a UniFi USW-24-POE over a VANDESAIL SFP+ DAC: the switch port stayed Disconnected under every negotiation setting on both sides. After ordering a 10Gtek cable the host reported network driver…
The poster could read and write OEM SFP variants with a REVELPROG and SFP adapter but not three Finisar references, suspecting manufacturer password protection. Finisar only documents the MSA user-EEPROM password in datasheets, not the proprietary manufacturer password, and the same happened with…
After upgrading OPNsense from the 21.x series to 22.7 a 10G SFP+ port stopped working with interface is disabled due to unsupported sfp; the newer Intel driver actively disables unrecognised modules. The poster tried to set ixgbe.allowunsupportedsfp=1 as a system tunable, which was not accepted…
A used IBM Emulex dual-port 10GbE SFP+ PCIe card (bent bracket, loose heatsink) made both TrueNAS Core and a Windows 10 PC hang for a long time during boot; the switch showed a link but neither OS saw the interface as active. Hardware involved: IBM Emulex dual-port 10GbE SFP+, Chelsio NICs, Intel…
On an Alta Labs Route 10 the SFP+ ports recognised FS.com SFP-H10GB-CU2M and 10Gtek CAB-10GSFP-P2M DACs (which had worked on MikroTik CRS devices) but no link came up against a MikroTik CRS309-1G-8S+; forcing 1G baseT full gave a 1 Gbps link only. Staff requested diagnostics and ethtool -I output…
A Chelsio CC2-S320E-SR (110-1114-30) dual-port SFP+ card recognised by TrueNAS would not link to a Dell PowerConnect 6224 through a QSFPTEK 0.5 m passive DAC: no link lights and status: no carrier, even after assigning IPs to cxgb0/cxgb1. Looping the DAC between two switch ports worked, and looping…
On an inexpensive firewall appliance (Intel Atom C3758R, six 1GbE igc ports and two SFP+ cages wired to the CPU) OPNsense showed four ix interfaces instead of two and the SFP+ ports stayed red/down with a copper SFP+ RJ45 module, although the link LEDs lit. ifconfig -v identified ix0/ix1 as the…
On a Lanner NCA-1510 the four Intel X553 ports (ix0-ix3) worked, but two Intel I210 Gigabit Fiber Network Connection interfaces (device 0x15368086) with SFP cages were absent in OPNsense even after loading ifigb via tunables; pfSense detected them by default. Another user had the same with a…
