hanahmily commented on PR #1198:
URL:
https://github.com/apache/skywalking-banyandb/pull/1198#issuecomment-4841478184
Thanks @copilot — addressed 4 of the 5 in `37284ce1` (CI green):
- **#1 `BatchWaitSec` truncation** — `schemaserver` now derives the index
batch wait via `batchWaitSeconds()` (ceil-to-seconds, floored at 1), so a
sub-second `--schema-server-flush-timeout` yields ≥1s instead of 0.
- **#2 `enabled_events` ignored** — `reconcilePipeline` now installs merge
samplers only when `PIPELINE_EVENT_MERGE` is enabled (empty list defaults to
MERGE for back-compat); added `TestReconcilePipeline_MergeEventNotEnabled`
locking the FINALIZE-only → no-samplers behavior.
- **#3 `sampler_load_failed` cardinality** — the `reason` label now maps to
a small, stable set of codes (`abi_mismatch`, `path_escape`, `open_failed`,
`ctor_failed`, `symbol_missing`, `config_marshal`, `config_invalid`, `panic`,
`load_error`); the full error is still logged.
- **#4 hard-coded `abi_version == 1`** — the validator now compares against
`sdk.ABIVersion` (single source of truth).
**#5 `TracePipelineRegistryService` shape** — leaving as-is for now. This
service is currently unwired (no `Register…Server` call) and is an intentional
planned-API stub (the doc comment references §2.3/§2.4 of the
schema-consistency design). This PR delivers config via the group-scoped
`common.v1.Group.pipeline` field over the existing `KindGroup` watch rather
than this RPC, so reconciling the Create/Update (name-less) vs Get/Delete
(`{group}/{name}`) shape is a deliberate API decision better made when/if the
RPC is actually implemented. Happy to follow up separately if maintainers
prefer to either drop the service or make it group-scoped now.
--
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]
