Complete replication requests for user management and commits #240
Conversation
farost
changed the title
| message SchemaCommit { | ||
| string name = 1; | ||
| bytes commit_record = 2; | ||
| bool profile_enabled = 3; |
There was a problem hiding this comment.
in core, how do we make commit profile to be collected? do we configure a profiling flag in the CLI at bootup time?
There was a problem hiding this comment.
We only commit from transactions, and, in transactions, their profiles are configured using tracing::enabled!(Level::TRACE)) <-- so it depends on the logging level.
There was a problem hiding this comment.
what happens if an admin creates a three node cluster, all with differing log levels? i'm not sure how, but it feels like something that we have to prevent right?
to highlight the meta theme: all servers in a cluster should be configured identically.
There was a problem hiding this comment.
I don't think they have to. We return the commit profile only of a single local execution, on the target node only. It's described in the server prs. I see no issue with different configurations. What do you see as an issue?
2 participants
Release notes: usage and product changes
Introduce protocol changes required by https://github.com/typedb/typedb-cluster/pull/642 :
Implementation