report: enable experiment run with local introspector #305
Conversation
Signed-off-by: David Korczynski <david@adalogics.com>
There was a problem hiding this comment.
Thanks @DavidKorczynski!
I left some comments below on how your code can better help with current experiment setting.
Please let me know if I missed anything : )
| @@ -0,0 +1,70 @@ | |||
| #!/bin/bash | |||
There was a problem hiding this comment.
Thanks so much for working on this promptly.
Could you please merge this into the existing docker_run.sh script?
I reckon they have a lot in common, and it's easier to adopt a local introspection for all experiments in that way.
Besides, maintaining one script is easier than two : )
Alternatively, it can be an independent script to set up the database (and do not care about running experiments), like upload_report.sh.
BTW, if you think modifying the code in docker_run.sh can simplify the integration (e.g., shorten the code, convert it into Python, etc.) or make the code more consistent, please feel free to do so.
The following comments will be based on the assumption of integrating this script into docker_run.sh.
| # limitations under the License. | ||
|
|
||
| # Comma separated project list | ||
| PROJECTS=${TARGET_PROJECTS} |
There was a problem hiding this comment.
The projects are the files in $BENCHMARK_SET (in docker_run.sh).
nit: Could you please use "${TARGET_PROJECTS}" to avoid unintended splitting or globbing?
Thanks, same below (e.g., $PWD -> "$PWD").
| echo $WORK_DIR | ||
| git clone https://github.com/ossf/fuzz-introspector $ROOT_FI | ||
| cd $ROOT_FI | ||
| python3 -m pip install -r ./requirements.txt |
There was a problem hiding this comment.
a minor note: docker_run.sh uses its own venv, please avoid conflict when merging : )
| cd $ROOT_FI/tools/web-fuzzing-introspection/app/ | ||
| FUZZ_INTROSPECTOR_SHUTDOWN=1 python3 ./main.py >> /dev/null & | ||
|
|
||
| SECONDS=5 |
There was a problem hiding this comment.
nit: IIRC, you mentioned that preparing the database for many projects (e.g., a full experiments with 200+ projects) can take a long time.
Would it be a good idea to make the waiting time depend on the project number? E.g. project_num * 5 seconds.
This can avoid spamming the log with "waiting messages".
| while true | ||
| do | ||
| # Checking if exists | ||
| MSG=$(curl -v --silent 127.0.0.1:8080 2>&1 | grep "Fuzzing" | wc -l) |
There was a problem hiding this comment.
This happens to be the same port for experiment result report.
| -g ${BENCHMARK_HEURISTICS} \ | ||
| -gp ${PROJECTS} \ | ||
| -gm 6 \ | ||
| -e http://127.0.0.1:8080/api |
There was a problem hiding this comment.
I reckon this can be merged into here.
|
Superseded by #423 |
Script for running an experiment using a local version of Fuzz Introspector's webapp.
For reference, there is also a more end-to-end version of this, which will run both the harness itself and the building of the introspector reports here. The logic in this PR is focused on using a local version of Fuzz Introspector's webapp such that all requests to
../api/..happens to the local webapp.Sample run: