jonathanmetzman
jonathanmetzman
Fuzzbench uses service account keys because when we didn't use them there was a horrible bug logging which is bad for 2 reasons: 1. It breaks experiments because we log...
- [ ] Give researchers access to buckets (this will allow them to get some more real time feedback. - [ ] Make more logs public. This also improves the...
It's only needed for fuzzer integration. Instead the integration script can just clone OSS-Fuzz.
This would give earlier feedback to our users than the current approach of waiting for a report to appear. This is particularly true for private experiments.
During the halloweeen experiment and sometime a few months ago (when @inferno-chromium upgraded the size of the db instance), we had to manually intervene in the db because Fuzzbench could...
This issue tracks switching to the queue based measurer. This will make measuring horizontally scalable, making experiments shorter, cheaper and results more realtime. It is also necessary for the new...
This is bad because it makes noop builds take two minutes (installing all packages we need takes 1:30).
We should have some documentation on how to create/do migrations for developers. We should also have some documentation on how to run them for users of fuzzbench. (Maybe fuzzbench should...
New experiments will be announced on this issue as discussed on #205. I think this is a good place to discuss experiments as well.
We have a benchmark now, matio, that has many crashes. For crash-based benchmarking we need a way for FuzzBench to identify when a fuzzer has found a crash. I propose...