Description of problem:

In Logging 6.2.6, the value of vector_buffer_max_byte_size is always 0.
This value is incorrect.
In another environment using Logging 6.2.2, the value is correct (268435488).
This looks like a regression in 6.2.6.

Version-Release number of selected component (if applicable):

OpenShift v4.20.0
cluster-logging.v6.2.6

How reproducible:

Always

Steps to Reproduce:

Step1. Install Logging 6.2.6 and configure ClusterLogForwarder.
Step2. Check the value of vector_buffer_max_byte_size using Prometheus query.

Actual results:

The value of vector_buffer_max_byte_size is 0 in Logging 6.2.6.

In Logging 6.2.2, the value was 268435488.

Expected results:

The value should be the same as in Logging 6.2.2 (268435488), which matches the max_size defined in vector.toml.

Additional information:

We haven't investigated deeply yet, but we noticed 2 points that may be related to this issue:
Point 1:
Why was as u64 omitted for buffer_max_byte_size?

https://github.com/ViaQ/vector/pull/219/files#diff-cd5d7e9d7dc7fc3347ca77c5322795f9115cfc8d0a71375d91de45d9d810ff51R54

as u64 is used for max_size_events:

https://github.com/ViaQ/vector/pull/219/files#diff-cd5d7e9d7dc7fc3347ca77c5322795f9115cfc8d0a71375d91de45d9d810ff51R50

Also, vector_buffer_max_event_size still works even with 6.2.6.
Maybe as u64 is important.

Point 2:
There is a commit that adds buffer_id handling for metrics:

https://github.com/vparfonov/vector/commit/72428f3954ace3f678830438579ad11c9a1ff1bf

We think similar handling may be needed for buffer_max_byte_size and buffer_max_event_size.
In upstream, buffer_id handling was added in a similar place:

https://github.com/vectordotdev/vector/blob/62e34462c4219a5a6e2497b08f7d91e2cb0b082b/lib/vector-buffers/src/internal_events.rs#L30