I have verified this is impacting all release builders for M65. 
Example:
https://uberchromegw.corp.google.com/i/chromeos_release/builders/lakitu-release%20release-R65-10323.B

In the R65 branch, the 9999 ebuild doesn't exist (just like the
error message says).  The file _does_ exist starting in the R66
branch.

I believe fizz is new enough that its failure to build in R65 isn't
an error.  What _is_ an error is that building lakitu (or any other
board except for fizz) should depend on fizz in this way.

I believe that there've been recent changes to `cros tryjob` that
might be relevant.

So... Passing to dgarrett@ to answer the question "is this a new bug,
or some longstanding problem that we've just discovered?"  And also
"who owns bugs in `cros tryjob`?".

As mentioned this is impacting release builders not only cros tryjob.
See:
https://uberchromegw.corp.google.com/i/chromeos_release/builders/lakitu-release%20release-R65-10323.B

This isn't a "cros tryjob" bug.

It sounds like some sort of ebuild dependency problem, but could be an uprev bug.

The fact R65 was building and is now broken certainly sounds like an ebuild dependency problem, which is a product bug.

I reverted a CL, https://chrome-internal-review.googlesource.com/#/c/chromeos/overlays/baseboard-fizz-private/+/589961/, which caused build breakage and this essentially removed the chromeos-config-bsp-baseboard package, which includes 9999 ebuild file, which the build is looking for now.

I am checking with others to see how to fix it.

Thanks.

OK.  I'm not sure what component to use, but like dgarrett@ said,
this is a product bug, not Infra.

Adding in the sheriffs, in case this needs further triage.

Thanks to Bernie to chumped in the change and restarted the builds. The builds look to be passing uprev stage now.

Verified with a lakitu-release tryjob on R65