docs: discussion of different kubernetes controller options. #5327
Conversation
| discovery.kubernetes "endpoints" { | ||
| role = "endpoint" | ||
| selectors { | ||
| role = "pod" |
There was a problem hiding this comment.
This is already documented in the reference docs, so not sure if it's worth calling out again here, but I imagine the two role arguments could be a bit confusing to newer folks?
Maybe a sentence like:
The top level `role` argument specifies the type of Kubernetes resource to query, and the selectors' `role` specifies the role of the selector.|
|
||
| There are several downsides to a `DaemonSet` which need to be considered: | ||
|
|
||
| - You cannot attach Persistent Volumes to `DaemonSet` pods. This means things like the Write-Ahead Log will use the Node's local disk, which may not be desired in shared environments. |
There was a problem hiding this comment.
Depending on the audience's expected familiarity, it might be good to note an example of why it wouldn't be desired (e.g. noisy neighbor; if any other pods on the node really need the disk there could be contention, slow, etc...)?
|
This seems to cover some common ground with #5325. Should we link the two documents or even merge them into one? cc @clayton-cornell |
Looking at the topics as they are in Draft state... I'd keep the separate for now.. but limit #5325 to defining the concepts/pros/cons and use this one to detail how to do things? That's kind of how it is now... with this PR containing the River configs. Better not to cram everything into a one-size-fits-all topic :-) |
Co-authored-by: Mischa Thompson <[email protected]>
|
Too much overlap with existing deploy page. I will propose additions on top of that, as well as the "Distribute Prometheus metrics scrape load |
github-actions
bot
added
the
frozen-due-to-age
PR Description
Which issue(s) this PR fixes
Notes to the Reviewer
PR Checklist