The bladeRF-cli -L option writes the FPGA image to SPI flash. This operation (if I remember correctly?) interrupts the operation of the FPGA and could be what's causing your issue, especially if you didn't do a power cycle after running the -L command.
Now, having said that, I do
not recommend using the -L command unless you are operating the bladeRF without a host machine (i.e. 'headless'). The flash operation takes a long time, not only to write the data to flash, but also to read it out during FPGA configuration after power-up. It's significantly faster to use
host-based auto loading. Host-based auto loading also makes it easier to keep the FPGA up-to-date as new versions are released. If you decide to use host-based auto loading, remember to clear the SPI flash image using the command: "bladeRF-cli -L X" otherwise you may run into race conditions or other issues.
If you're certain you want the FPGA image stored in SPI flash, then only issue the command once to write the image to flash. Then power cycle the board and wait until all four LEDs turn on before opening the bladeRF with "bladeRF-cli -i". I also recommend upgrading the FX3 firmware to v2.3.1 if you haven't already. This version fixes an issue that interrupted FPGA configuration if the host opens the bladeRF before the FPGA is fully configured.
If you're still experiencing issues, please run the CLI in verbose mode so we can get a better understanding of where it's failing in the code. To do this add "-v verbose" to bladeRF-cli.