wu-sheng opened a new pull request, #13904:
URL: https://github.com/apache/skywalking/pull/13904

   ### Surface BanyanDB queue batch / message granularity metrics (BanyanDB 
#1169)
   
   Follow-up to the SWIP-15 BanyanDB self-observability redesign. [BanyanDB 
#1169](https://github.com/apache/skywalking-banyandb/pull/1169) added per-batch 
and per-message catalogs to the batch-write queue metrics; this PR surfaces the 
subset that genuinely populates, modeled per the cluster / container / group 
scopes.
   
   - [x] If this is non-trivial feature, paste the links/URLs to the design doc.
     - Design: 
[SWIP-15](https://github.com/apache/skywalking/blob/master/docs/en/swip/SWIP-15.md)
   - [x] Update the documentation to include this new feature.
     - `docs/en/banyandb/dashboards-banyandb.md` — added the new metric rows 
under the Liaison / Data / Endpoint tables.
   - [x] Tests (including UT, IT, E2E) are added to verify the new feature.
     - E2E assertions added to the no-FODC so11y case 
(`test/e2e-v2/cases/banyandb/banyandb-cases.yaml`); MAL fixture coverage 
regenerated. Verified locally: e2e **20/20 passed** on a clean OAP image 
against a BanyanDB #1169 cluster (liaison + hot data); 
`MALExpressionExecutionTest` 0 banyandb failures.
   
   #### New metrics (5 — all e2e-verified to return data)
   
   | Metric | Scope | Source family |
   |---|---|---|
   | `publish_batch_throughput` | instance (liaison) | 
`banyandb_queue_pub_total_batch_finished` |
   | `publish_batch_latency_p99` | instance (liaison) | 
`banyandb_queue_pub_total_batch_latency` |
   | `queue_sub_message_throughput` | instance (data) | 
`banyandb_queue_sub_total_message_finished` |
   | `queue_batch_throughput` | endpoint (per group) | 
`banyandb_queue_sub_total_batch_finished` |
   | `queue_message_throughput` | endpoint (per group) | 
`banyandb_queue_sub_total_message_finished` |
   
   #### Intentionally not modeled (verified empty-by-design)
   
   Confirmed by probing a live #1169 cluster and reading the BanyanDB source:
   
   - `queue_sub_batch_throughput` / `queue_sub_batch_latency_p99` at 
instance/data scope — a data node ingests via the per-message dispatch path, 
which ticks only the message counter; the sub-side batch counters fire on the 
liaison's OAP-facing `handleEOF` hop, so they are always empty on a data node.
   - `queue_batch_latency_p99` at endpoint scope — 
`banyandb_queue_pub_total_batch_latency` is emitted with `group=""`, so it 
cannot roll up per Endpoint (it is meaningful only at instance scope, where 
`publish_batch_latency_p99` covers it).
   
   #### Other changes
   
   - `test/e2e-v2/script/env`: bump `SW_BANYANDB_COMMIT` `#1167` → `#1169` (the 
merge commit that ships these families). The `#1167..#1169` delta is docs + the 
additive metrics PR + an off-path lifecycle/dump refactor; the full write/query 
path is exercised green by the so11y e2e, and CI runs the full banyandb e2e 
matrix.
   
   - [x] Update the [`CHANGES` 
log](https://github.com/apache/skywalking/blob/master/docs/en/changes/changes.md).


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to