-
Hi I am testing a range of local LLM apps and Flowise got a lot of upvotes on reddit when I asked about a good tool for local+RAG and configurability. But all these tools all want a heck of a lot of dependences and I am wary of breaking some/all when I install a new one. So: Background - Any reason for global install? I really don't want to break something else I am using if Flowise installs something that conflicts. Will it work if I did this instead? I see there is also a docker. I suppose I could run that, but I've already had problems with another local LLM tool (Cheshire Cat) where because the app is running in docker, but needs to connect to LLMs outside docker, it's a pain and a hassle. So I would prefer not to use docker, but also not to install a bunch of things globally if I can avoid it. As an aside, my node --version also shows 21.1.0, is this also OK? Thanks |
Beta Was this translation helpful? Give feedback.
Replies: 4 comments 1 reply
-
Bump to this |
Beta Was this translation helpful? Give feedback.
-
You would use -g when you are going to be using Flowise as a service running on a server. For example, you're putting Flowise on a server or docker container that's going to run your staging or production flows. If you're doing dev to make changes to Flowise source code, you wouldn't install at all. Clone the repo and run local. Omitting the -g option on an npm install would install flowise as a dependency in another project (you would need to have that project already set up). So, unless you're making a nodejs project that's going to be extending Flowise, use the -g option so that Flowise is installed globally and can be started as a system service on your server. Hope that helps. |
Beta Was this translation helpful? Give feedback.
-
Thanks for the explanation. It would not be running on a server, but neither do I really want to change it or do dev. So I am wary of installing it globally, as I'm testing a range of other things and don't want to cause any conflicts by dumping a range of requirements outside a conda or other environment, that's all. |
Beta Was this translation helpful? Give feedback.
-
Under that scenario running it using docker makes the most sense since that's the exact scenario docker was created for in the first place. It would still be a -g install on the docker container. The Flowise source has a docker compose file that will get things going quickly. Good luck. |
Beta Was this translation helpful? Give feedback.
You would use -g when you are going to be using Flowise as a service running on a server. For example, you're putting Flowise on a server or docker container that's going to run your staging or production flows.
If you're doing dev to make changes to Flowise source code, you wouldn't install at all. Clone the repo and run local.
Omitting the -g option on an npm install would install flowise as a dependency in another project (you would need to have that project already set up). So, unless you're making a nodejs project that's going to be extending Flowise, use the -g option so that Flowise is installed globally and can be started as a system service on your server. Hope that helps.