Fuzzing not happening for the past week

[alexcrichton]

I was taking a look at the fuzzer statistics for the Wasmtime project today and noticed that we haven't had any fuzzing since October 15. I wanted to raise this in case it wasn't already known, but if this is a known issue I'm happy to close!

[lum1n0us]

I noticed the same problem in the WAMR project. There is no test data for recent 7 days although all build status are green.

[nathaniel-brough]

I also see the same problem with the following projects;

• fastfloat,
• liboqs,
• halide
• quantlib
• simdutf
• surrealdb
• yoga

I don't see the same problem with;

• gitoxide
• rhai
• starlark-rust
• tinyusb
• tokio
• typst

There doesn't seem to be any pattern between which projects are working and which aren't. Most of the above projects all having working builds so I'm unsure what is going on here.

[ktmf01]

Same goes for flac, although it seems some of the fuzzers did run after Oct 15. I looked at day-to-day statistics for libfuzzer_asan_flac

• On Oct 15 all fuzzers ran, each had between 20 and 50 hours of runtime
• On Oct 16, only 3 fuzzers, each had between 3 and 6 hours of runtime
• On Oct 17, only 2 fuzzers (none of them the same as Oct 16), both 1.8 hours of runtime
• On Oct 18, no statistics available (I assume no fuzzers ran)
• On Oct 19, 4 fuzzers ran for between 3 and 7 hours
• On Oct 20, no statistics
• On Oct 21, 2 fuzzers ran, both 1.8 hours
• On Oct 22, no statistics

So, it seems the frequency with which fuzzers have started has dropped dramatically?

[oliverchang]

Sorry for the delay. We're looking into the root cause of the reduced fuzzing capacity across OSS-Fuzz.

[jonathanmetzman]

This should be improving but isn't fixed yet.

[alexcrichton]

I can confirm that at least for Wasmtime we had 15 hrs of fuzzing happen on 2024-10-28 (yesterday)

[lum1n0us]

I also confirm that we have received the data for the WAMR project from October 31st. Thank you.

[nathaniel-brough]

All the projects I've checked look fine now too.

[nathaniel-brough]

Actually nvm. Looks like liboqs hasn't seen any fuzzing since Nov 2. So as @jonathanmetzman mentioned it's improved, but not fixed.