Stop copying truss server code over for "trussless" #1189
Conversation
@dsingal0 can you try context image "v0.9.44-rc" to verify it can resolve the issue? |
|
@tianshuc0731 I also think that we might need to merge #1187 to properly test dhruv's changes. it lgtm |
…r-stop-copying-truss
Merged the change, @dsingal0 please test this new image "v0.9.44-rc2" |
| @@ -106,7 +108,7 @@ RUN mkdir -p {{ supervisor_log_dir }} | |||
| COPY supervisord.conf {{ supervisor_config_path }} | |||
| ENV SUPERVISOR_SERVER_URL="{{ supervisor_server_url }}" | |||
| ENV SERVER_START_CMD="supervisord -c {{ supervisor_config_path }}" | |||
| ENTRYPOINT ["/usr/local/bin/supervisord", "-c", "{{ supervisor_config_path }}"] | |||
| ENTRYPOINT ["supervisord", "-c", "{{ supervisor_config_path }}"] | |||
There was a problem hiding this comment.
one non-blocking comment, since this resulted in an actual bug/issue, would be awesome if we could also have an integration test for this, don't know how easy that would be to put together
There was a problem hiding this comment.
No easy way on top of my mind, maybe what we want is to have a different testing docker image which installs supervisord in a different path? Any better ideas?
There was a problem hiding this comment.
yeah maybe something like that. no need to block on this then
🚀 What
We caught some bugs in trussless custom server in real life when customers try to start their server using
python main.pyas start commandhttps://basetenlabs.slack.com/archives/C037X7HFAE4/p1728672357619419?thread_ts=1728574438.184139&cid=C037X7HFAE4
💻 How
python main.pyto start fastapi server, so we can catch similar type of bugs in the future🔬 Testing
Passed integration test
https://github.com/basetenlabs/truss/actions/runs/11353710099