Single mode, single strand fiber handles 1Gbit but not 10Gbit

Question

I have a strand of dark fiber between two buildings, with the run being roughly 1.5 miles/2.4km, and it is currently operating with 1Gb BX 10km optics, but when I try to put in 10Gb BX 10km optics, I don't get link. I have tried two separate pairs of the 10Gb optics that were fresh out of the box. I know both switches are capable of 10Gb, and both are in auto negotiate. Is there anything in the fiber run itself that may stop it from doing 10Gb but allow 1Gb? The 10Gb optics I'm using are FiberStore Generic SFP-10G-BX 10KM 1270/1330 (fs.com/products/74681.html and fs.com/products/74682.html).

EDIT: One switch is a Cisco 9396PX with port config as below:

interface Ethernet1/9
switchport mode trunk
switchport trunk
allowed vlan 98,899

The other switch is a Ubiquiti Edgeswitch 16XG with port config below:

interface  0/1
vlan participation exclude 1
vlan participation include 98,899
vlan tagging 98,899
exit

Answer 1

1. Try to link the pair of transceivers with a simple patch cable - not all of them are compatible and the wavelengths need to match (crossed over). There's no offical standard for 10GBASE-BR (as it should be called due to using -R PCS code), so there are some degrees of freedom. (The fs.com SFPs you've linked to should work together, obviously, but possibly one's DOA.)
2. Double check whether the SFP+ modules are compatible with your switch. Many switches only accept "original" modules or ones pretending to be. You might need to get vendor-compatible versions.
3. Check the switch logs for error or incompatibility messages.
4. Check the transceiver diagnostics (see Jeff's comment) - it should show at least some receive power when plugged together.
5. With a low-power signal or significant error counts on the ports, check the fiber for overbending (< 15 mm radius) and the terminals for dirt or damage (requires a scope). It could also be a faulty splice.

On marginal fiber it's theoretically possible for 10G to fail while 1G barely works but that's very unlikely, especially on such a short run.

A wavelength problem can be ruled out pretty much: 1000BASE-BX10 uses 1310 and 1490 nm while (most) 10GBASE-BR transceivers use 1270/1330 nm - unless of course, there's a WDM multiplexer somewhere, splitting off 1270 nm.

My guess is that it's #2.