Page History

...

Pegasus supports ccache to accelerate compilation. It significantly reduces the building time, by caching the intermediate results on your disk. We highly recommend this tool for Pegasus development.

sudo apt install ccache

ccache -M 5G # configure the cache size

...

Pegasus runs a Github Actions workflow/TravisCI for each of the submitted pull-request. The CI process procedure checks:

• If your PR title matches conventional commit
• If your code is well-formatted
• If your changes can pass all the tests. You can run `./run.sh test` on your local environment.

For more details about how we run Github Actions, please refer to this doc: Github Actions .

In addition to testing-per-PR, we also run daily workflows at https://github.com/pegasus-kv/pegasus-docker that check:

• If rdsn/pegasus can be built upon various platforms and compilers.
• If all code is sanitized (address/leak/thread/undefined).

Heap Profiling

By default, Pegasus enables gperftools and utilizes tcmalloc for memory allocation. To analyzing the memory/cpu usage of Pegasus server, you can run pprof against the target server. The Pegasus server will retrieve profiling data via gperftools library and respond via HTTP. 

pprof --svg http://127.0.0.1:34801/pprof/heap > heap.svg

Thrift

Pegasus uses thrift as the serialization tool for RPC data. To modify the RPC structs in Pegasus, or to add a new RPC type, you need to update the corresponding ".thrift" protocol file, and run:

./rdsn/compile_thrift.py # if the thrift is in rdsn

./src/idl/recompile_thrift.sh # if the thrift is in pegasus

We directly manage thrift as one of our third-parties, that is, it requires no separate installation. After running `./run.sh build` it will be automatically installed under `./rdsn/thirdparty/output/bin`.