Deployment and TaskQueue Stats API #571
Conversation
| repeated temporal.api.taskqueue.v1.TaskQueueStatDetails per_queue_stats = 1; | ||
|
|
||
| // Aggregated backlog count across all task queues in this deployment version. | ||
| int64 approximate_total_backlog_count = 2; |
There was a problem hiding this comment.
Name is derived from TaskQueueStatDetails's approximate_backlog_count field.
stephanos
force-pushed
the
task-stats
branch
8 times, most recently
from
6b25fed to
fa2df53
Compare
stephanos
force-pushed
the
task-stats
branch
4 times, most recently
from
f46f542 to
9778bd9
Compare
stephanos
changed the title
|
|
||
| // Optional. Deployment version to return stats for. | ||
| // If left empty, the current version will be queried. | ||
| string deployment_version = 3; |
There was a problem hiding this comment.
Need to double-check that build_id isn't a better field name here.
There was a problem hiding this comment.
We're in the middle of that discussion right now in the Versioning crew, we have not fully converged yet. If you want to merge this asap, I think build_id is the safer alternative.
| message TaskQueueStatsInfo { | ||
| temporal.api.taskqueue.v1.TaskQueue task_queue = 1; | ||
| temporal.api.enums.v1.TaskQueueType task_queue_type = 2; | ||
| temporal.api.taskqueue.v1.TaskQueueStats task_queue_stats = 3; | ||
| } |
There was a problem hiding this comment.
This message only exists because TaskQueueStats doesn't contain the task queue name and task_queue_type.
I considered adding them to the existing TaskQueueStats, but was discouraged to do so.
stephanos
force-pushed
the
task-stats
branch
2 times, most recently
from
ba4700d to
2c3d3cf
Compare
| // GetWorkerDeploymentStats returns the stats for all task queues in a deployment version. | ||
| rpc GetWorkerDeploymentStats (GetWorkerDeploymentStatsRequest) returns (GetWorkerDeploymentStatsResponse) { | ||
| option (google.api.http) = { | ||
| get: "/namespaces/{namespace}/worker-deployments/{deployment_name}/{deployment_version}/stats" |
There was a problem hiding this comment.
Since deployment_version can be empty, I'm not sure this is right yet.
There was a problem hiding this comment.
I think we should only put the name in the url.
There was a problem hiding this comment.
This shows I'm not familiar with how the grpc-http bridge works. I suppose the other args are passed as query parameters?
| // GetTaskQueueStats returns stats for a single task queue. | ||
| rpc GetTaskQueueStats (GetTaskQueueStatsRequest) returns (GetTaskQueueStatsResponse) { | ||
| option (google.api.http) = { | ||
| post: "/namespaces/{namespace}/task-queues/{task_queue.name}/type/{task_queue_type}/get-stats" |
There was a problem hiding this comment.
Does this work with the task_queue_type enum in there?
There was a problem hiding this comment.
Copilot says we can and caller would put the numeric values of the enum in the URL.
There was a problem hiding this comment.
Hm, I don't know if the numeric enum representation is a good choice here.
stephanos
changed the title
READ BEFORE MERGING: All PRs require approval by both Server AND SDK teams before merging! This is why the number of required approvals is "2" and not "1"--two reviewers from the same team is NOT sufficient. If your PR is not approved by someone in BOTH teams, it may be summarily reverted.
What changed?
Added two new APIs:
GetWorkerDeploymentStatsGetTaskQueueStatsWhy?
To return task queue stats to users for making scaling decisions.
Breaking changes
No.
Server PR
temporalio/temporal#7581 (feature complete, but not yet production-ready)