PrivatePuffin
commented
2 years ago I like the idea, but I think this should be something you should request from iX-Systems to handle, as they create/design the general architecture.
Adding a servicemesh on top of their architecture, is going to ask for a unworkable support burden. In addition, our current user target is only Home/Enthousiast users, which do not require servicemesh. Our future target would be SMB/SME with just a few servers in a cluster, setups which require servicemesh are not really something we see as usecase for those small clusters either.,
Also: SCALE does not support any clustering at the moment, so there is no use for it anyway. Adding servicemeshing for single-node clusters is simply silly.
TLDR: If you need servicemeshing, your deployment scale is basically too big for TrueNAS SCALE and if you want service meshing you most likely are better off completely DIYíng it yourself.
Hence i'm going to close this issues, as even if we're going to add it, it would be 2023 or 2024. We try to only keep issues open for things that are at least aimed to get worked on in a somewhat reasonable timeframe.
That being said If you come up with a nice design, PR's will always be reviewed fairly.
astro-stan
truecharts-admin
I was debating whether to file this as a "new app" or "feature or enhancement" request. Please attach appropriate labels if needed.
Is your feature request related to a problem? Please describe. I have been looking for a 'set-it-and-forget-it' solution that will allow me to monitor all of the ~apps~ pods running in the cluster, while also providing mTLS for service-to-service communication and other goodies.
I cannot think of a better solution than adding support for a service mesh. A service mesh provides: :heavy_check_mark: Automatic mTLS between services (issuing, rotation, the whole life cycle) :heavy_check_mark: Automatic request retries and/or timeouts :heavy_check_mark: Monitoring and tracing :heavy_check_mark: Fault injection (for simulating failures and observing how they affect your cluster) :heavy_check_mark: Traffic split (canaries, blue/green deploys) :heavy_check_mark: And more...
It does this by injecting a proxy sidecar container into every pod, which allows it to provide all of the above features without altering the application being run.
Describe the solution you'd like Adding support for Linkerd2 and/or Istio would be ideal. Both have advantages and disadvantages but the gist is:
If I had to choose one, I would say Linkerd2 is the better choice at this stage. Istio can be added later as a more advanced alternative if there is enough interest from the community.
Describe alternatives you've considered There is no suitable alternative to a service mesh as far as I am aware. Well, one can start manually implementing all of the features a service mesh provides, but this becomes harder and harder as the cluster scales or more apps are being added. Therefore, I don't believe this to be a realistic alternative.
Additional context DISCLAIMER I am not very familiar with the Truecharts architecture, and I have only just began looking at the common chart library. However, I am pretty certain that adding support for a service mesh it would not be as easy as adding a new app. It would most likely require changes to the common chart templates. I am available for discussions / clarifications should they be needed.