Configure operator's peering via centralised settings #572
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The legacy peering mode is now 1.5 years since it was deprecated in the early stages. There is no need to support it anymore, as barely anyone still use it in their clusters (taking into account that peering per se it not used often at all). But the legacy mode complicates the coming features with dynamic cluster scanning. So, the time has come to drop it — without releasing the major version of the framework.
Previously, the peering setup was stored in a pseudo-peer, so called "ourselves" — as detected at startup. This also included the "legacy peering mode" (which is now removed), and the whole peering configuration — namespace, peering object name, priority, our own identity, etc. Now, this is split: the peering settings are moved to settings; the dynamic runtime identity is used directly; so as the namespace (with potentially multiple namespaces for multiple peering tasks in mind). The `Peer` class is only left for the actual peers as read from the peering object, and solely for the purpose of doing the math with lifetime & deadlines.
With the default values, they override the default values defined in the settings object. Which is not a big deal until they match, but can be an issue if the configuration defaults are changed later. It is better to have one and only one source of truth for the default values. A downside: the defaults will not be seen in `--help`. But these values can be misleading anyway, as the operator can override the values in the startup handlers.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The settings include:
The values passed to
operator()/run()via kwargs or CLI options override the default settings. But the values set in the startup handlers override the kwargs/options.The essential change here is that we get rid of an pseudo-peer "ourselves" that is used to store the configuration of the current operator, confused with other
Peerobjects of other operators as detected at runtime. That was one class for two distinct purposes. It is no more.Internally, also extract the
Identitytype (a string) for operators' ids — so that they are not fixed with other strings by accident, and do some renames.This is an extraction from a bigger PR for the reactor refactoring for multi-dimensional cluster observation — to separate the semantically distinct changes and reduce the size of the PR. There might be no direct value in this change alone.