Skip to content

Shift ingestion feature metrics to after store write  #489

Closed

Description

@zhilingc

Is your feature request related to a problem? Please describe.
Existing metrics collection for ingestion is done in parallel to store writes:

               +
               |
       +-------+--------+
       |                |
+------+-----+   +------+-----+
|            |   |            |
|   store    |   |  metrics   |
|   write    |   |   write    |
|            |   |            |
+------------+   +------------+

While this does give a rough gauge as to what data looks like in flight, it doesn't actually reflect the status of the data in the store, which is ultimately what we are concerned about, since write failures are common.

Describe the solution you'd like
Ideally we want to write metrics POST store writes, as so:

              ...
               +
               |
         +-----+-----+
         |           |
         |   store   |
         |   write   |
         |           |
         +-----+-----+
               |
      +--------+--------+
      +                 +
   success           failure
      +                 +
      |                 |
+-----+-----+     +-----+-----+
|  write    |     |  write    |
|  sucess   |     |  failure  |
|  metrics  |     |  metrics  |
+-----------+     +-----------+

Metrics collected post store writes would more accurately reflect the data in the stores.

Activity

  1. added this to the v0.5.0 milestone on Mar 10, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    area/ingestionThe ingestion Beam component and storage-related itemskind/techdebt

    Type

    No type

    Projects

    No projects

      Milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions