Skip to content

Assign WorkerGroup via worker configuration #330

Closed

Description

@imbajin

Currently, Vermeer Workers can be assigned to multiple WorkerGroups through the administrator REST API, thereby achieving parallel scheduling in different graph spaces.

In addition to this method, we would like to implement that when a new Worker is registered in the Master, it can be directly assigned to the specified WorkerGroup through the WorkerGroup name specified in the Worker configuration file.

Activity

  1. changed the title [-]**Assigning WorkerGroup via worker configuration**[/-] [+]Assign WorkerGroup via worker configuration[/+] on May 30, 2025
  2. added theissue type on May 30, 2025
  3. Ethereal-O commented on May 30, 2025

    @Ethereal-O
    Contributor

    Hello everyone, I have briefly outlined the calling process and related concepts such as Space, Graph, Worker, WorkerGroup, etc( https://ethereal-o.github.io/2025/05/21/hugegraph-vermeer-reading/ ).
    Meanwhile, I also briefly outlined the plan for implementing this task:

    1. Worker loads the WorkerGroup configuration item from the configuration at startup.
    2. Pass configuration items to the Master when SayHello.
    3. The Master uses the passed WorkerGroup configuration item in the CreateWorker function to assign a Group to the Worker.

    At present, I think the points that need attention and solutions are as follows:

    • Default configuration: Worker passes in a "$" string.
    • Priority of recovery from object storage: The priority of recovery from object storage is higher, and I believe that ensuring the stability of the Master is the primary condition.
    • Worker trustworthiness: Always believe that the worker is trustworthy and faithfully executing the correct code.

    I would like to ask if my previous approach is correct? And are there any other areas that need attention? thank.

  4. hankwenyx commented on Jun 3, 2025

    @hankwenyx
    Collaborator

    Hello everyone, I have briefly outlined the calling process and related concepts such as Space, Graph, Worker, WorkerGroup, etc( https://ethereal-o.github.io/2025/05/21/hugegraph-vermeer-reading/ ). Meanwhile, I also briefly outlined the plan for implementing this task:

    1. Worker loads the WorkerGroup configuration item from the configuration at startup.
    2. Pass configuration items to the Master when SayHello.
    3. The Master uses the passed WorkerGroup configuration item in the CreateWorker function to assign a Group to the Worker.

    At present, I think the points that need attention and solutions are as follows:

    • Default configuration: Worker passes in a "$" string.
    • Priority of recovery from object storage: The priority of recovery from object storage is higher, and I believe that ensuring the stability of the Master is the primary condition.
    • Worker trustworthiness: Always believe that the worker is trustworthy and faithfully executing the correct code.

    I would like to ask if my previous approach is correct? And are there any other areas that need attention? thank.

    Your approach is correct. And your code reading notes are very good, which can be further improved in detail to provide assistance to developers who will participate in the future.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions