I'm following up instructions at
https://chromium.googlesource.com/infra/goma/server/ and
https://chromium.googlesource.com/infra/goma/client to set up Goma service within Intel. However, I met some issues to make it run well and I have some further questions that need your help.
So far, my progress is a server with Ubuntu 19.04 has been set up, while my client side machines can be Windows or Linux. If I run client on same machine as server (export GOMA_SERVER_HOST=localhost), it seemed Goma already worked well. But if I change GOMA_SERVER_HOST to real <server_hostname>, compiler_proxy couldn't start successfully using "goma_ctl.py start". You may see I met same issue for other client machines. The command for me to start remoteexec_proxy is "remoteexec_proxy --port 5050 --platform-container-image "docker://<id>" --remoteexec-addr <server_hostname> --remote-instance-name <server_hostname> --whitelisted-users <user>". So my detailed questions are:
1. What's the issue that localhost could work, but server_hostname for GOMA_SERVER_HOST couldn't work? I think I must misunderstand some options here, like --remoteexec-addr, --remote-instance-name, --service-account-json.
2. How to specify the option for "--whitelisted-users", if I have multiple users? My Windows machine is in a special domain, while server is not.
3. How can I set up the server with more than 1 machine? How should I change the configuration of client?
4. How can I restrict the users of Goma server?
If you have some more detailed document, or detailed description of some options for both server and client commands, or some further examples, that would be very helpful. Thanks!