Support grouping rules such that rules within a group are executed sequentially #1095
Comments
As an additional note, we'll also likely want to use such groups to indicate which rules are allowed to access remote storage. Generally remote storage should be avoided for reliability reasons, and only be used for human trending. |
Is there any timeline for this feature? |
It's not a super-trivial one so it will take a bit. But it's high up on the list of next features to work on. |
Are there any thoughts on how these groupings will be expressed syntactically? It occurred to me that it can be inferred automatically by comparing (lhs) rule names+labelsets with (rhs) metric selectors, and grouping all possibly-matching rules together. However, this is a bit automagic, so it's not necessarily the obvious solution. I'm considering making an unpublished patch to implement this, just to stay on top of mounting technical debt. I would prefer not to diverge too far from whatever plans you have without good reason, so I'm asking what those plans are. |
A proposal for fixing this with backward compatibility is here: https://docs.google.com/document/d/1AgRT1sJyvxwVx6ZLffUKFZ3yYFt01LSz_ohLQbA2yc4/edit |
Added in 2.0. |
This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs. |
Introduce a mechanism to wrap a set of rules into a group in such a way that all rules within that group are executed sequentially (vs. in parallel). This allows dependent rules to work deterministically.
Historical note: See also discussions on #1088 and #17.
The text was updated successfully, but these errors were encountered: