Files
milvus/pkg/metrics/metrics.go
T
0d2637fd63 fix: keep proxy metric status label stable, split cause into its own label (#51495)
issue: #51493

### What

#50221 split the `status` label values of `milvus_proxy_req_count` /
`milvus_proxy_grpc_latency` in place. Released in **v2.6.19**, that
silently zeroes every existing dashboard panel and alert rule matching
`status="fail"` / `status="rejected"` — an alert that stops firing
rather than erroring.

This keeps the classification but moves it onto its own dimension,
restoring the status domain to what <= v2.6.18 emitted:

```
status: success | fail | rejected | retry | total | abandon    (as before v2.6.19)
cause:  user | system | cancel | na                            (new)
```

| v2.6.19 / v2.6.20 | this PR |
|---|---|
| `status="fail_input"` | `status="fail", cause="user"` |
| `status="fail_system"` | `status="fail", cause="system"` |
| `status="rejected_user"` | `status="rejected", cause="user"` |
| `status="rejected_system"` | `status="rejected", cause="system"` |
| `status="cancel"` | `status="fail"` or `"rejected"`, `cause="cancel"`
|
| `success` / `retry` / `total` / `abandon` | unchanged, `cause="na"` |

### Why a label instead of new status values

- **Old queries work unchanged and count each request once.**
`status="fail"` is again every hard failure; Prometheus aggregates over
`cause` for free. That rollup is what a label dimension *is* — the split
forces every consumer who just wants "how many failures" to hand-write
`status=~"fail_.*"`.
- **Cardinality is unchanged.** `cause` is functionally dependent on the
outcome, so the realized `(status, cause)` pairs are exactly the series
the split already produces. Additive, not multiplicative.
- **New consumers lose nothing**: alert on `status="fail",
cause="system"`; route `cause="user"` to the owning application team.

The alternative — emitting old and new `status` values side by side
during a transition — was rejected: it defers the break rather than
removing it (the old values must still be dropped eventually, forcing
the same migration later), and meanwhile double-counts every request in
unfiltered aggregations.

`cancel` is folded into `cause` for the same reason: before v2.6.19
client cancellations were counted as `fail` (cancel in the response
status) or `rejected` (cancel at the interceptor), so promoting it to a
`status` value quietly makes `status="fail"` under-count versus older
releases. As a cause it stays excludable via `cause!="cancel"` without
redefining `status`.

### Also in this PR

- The 7 `snapshot_impl` sites that emitted a bare `fail` with no
classification now report their real cause via `failMetricLabel(err)`
(e.g. `snapshot_metadata_uri is required` is `cause="user"`, not a
system fault). Their `status` is unchanged.
- `deployments/monitor/grafana/milvus-dashboard.json`: the "Faild
Request Rate" panel goes back to `status="fail"` — a verbatim revert of
what #50221 changed, which is the compatibility claim demonstrated.
- Dev docs and stale comments that named the old label values.

### Verification

- `TestParseMetricLabelStatusDomainIsStable` pins the status domain, so
re-splitting it fails CI rather than shipping.
- Adding a label makes every `WithLabelValues` site arity-sensitive **at
runtime, not compile time** — a missed site panics in production. Unit
tests do not reach all of them (coverage shows
`GetRestoreSnapshotState`, `ListRestoreSnapshotJobs`, `PinSnapshotData`,
`UnpinSnapshotData` at 0%), so all **58 emit sites were verified
statically via AST**, not only the ones tests happen to hit.
- Passing: `pkg/metrics`, `pkg/util/requestutil`, `pkg/util/merr`,
`internal/distributed/proxy` (incl. `httpserver`, which covers the REST
emit path), and the `internal/proxy` snapshot / metric-label tests.
`golangci-lint` clean on every touched package.
- Pre-existing and unrelated: `service_test.go` bind failures (`listen
tcp :19529: address already in use`) reproduce identically on unmodified
master — a local process holds the port.

### Rollout

Should be cherry-picked to 2.6 so the window in which the fine-grained
`status` values exist stays confined to v2.6.19–v2.6.20. Users who
already adopted the new values on those two releases need a one-time
query change (`status="fail_system"` -> `status="fail",
cause="system"`); that is a known, bounded set, and preferable to
leaving every pre-2.6.19 dashboard silently broken.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Signed-off-by: zhenshan.cao <zhenshan.cao@zilliz.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 17:18:40 +08:00

240 lines
8.3 KiB
Go

// Licensed to the LF AI & Data foundation under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package metrics
import (
// #nosec
_ "net/http/pprof"
"github.com/prometheus/client_golang/prometheus"
)
const (
milvusNamespace = "milvus"
AbandonLabel = "abandon"
SuccessLabel = "success"
FailLabel = "fail"
CancelLabel = "cancel"
TotalLabel = "total"
RetryLabel = "retry"
RejectedLabel = "rejected"
// Values of the "cause" label, a dimension orthogonal to "status" that names
// the responsible party for a failed request, so monitoring can tell a
// user-input error (the caller must fix the request) apart from an internal
// system error (operators must intervene). It is a separate label rather
// than extra "status" values on purpose: the coarse status stays a valid
// query on its own ("fail" is still every hard failure, aggregated over
// cause by Prometheus), so dashboards written against it keep working.
// Cause is functionally dependent on the outcome, so the realized
// (status, cause) pairs are additive, not the product of both domains.
CauseUser = "user" // the request itself is at fault: bad arguments, missing auth, no such collection
CauseSystem = "system" // Milvus is at fault: component failure, IO error, internal bug
CauseCancel = "cancel" // neither party: the client gave up before the request completed
CauseNA = "na" // no cause applies: the request did not hard-fail
HybridSearchLabel = "hybrid_search"
InsertLabel = "insert"
DeleteLabel = "delete"
UpsertLabel = "upsert"
SearchLabel = "search"
QueryLabel = "query"
UpsertQueryLabel = "upsert_query"
DeleteQueryLabel = "delete_query"
ReQueryLabel = "requery"
CacheHitLabel = "hit"
CacheMissLabel = "miss"
TimetickLabel = "timetick"
AllLabel = "all"
)
const (
PreferredNodeHitLabel = "hit"
PreferredNodeMissLabel = "miss"
PreferredNodeUnavailableLabel = "unavailable"
PreferredNodeRejectedLabel = "rejected"
)
const (
UnissuedIndexTaskLabel = "unissued"
InProgressIndexTaskLabel = "in-progress"
FinishedIndexTaskLabel = "finished"
FailedIndexTaskLabel = "failed"
RecycledIndexTaskLabel = "recycled"
// Note: below must matchcommonpb.SegmentState_name fields.
SealedSegmentLabel = "Sealed"
GrowingSegmentLabel = "Growing"
FlushedSegmentLabel = "Flushed"
FlushingSegmentLabel = "Flushing"
DroppedSegmentLabel = "Dropped"
StreamingDataSourceLabel = "streaming"
BulkinsertDataSourceLabel = "bulkinsert"
CompactionDataSourceLabel = "compaction"
Leader = "OnLeader"
FromLeader = "FromLeader"
HookBefore = "before"
HookAfter = "after"
HookMock = "mock"
ReduceSegments = "segments"
ReduceShards = "shards"
BatchReduce = "batch_reduce"
FunctionChainLevelL0 = "l0"
Pending = "pending"
Executing = "executing"
Done = "done"
ImportStagePending = "pending"
ImportStagePreImport = "preimport"
ImportStageImport = "import"
ImportStageStats = "stats"
ImportStageBuildIndex = "build_index"
ImportStageWaitL0Import = "wait_l0_import"
compactionTypeLabelName = "compaction_type"
isVectorFieldLabelName = "is_vector_field"
segmentPruneLabelName = "segment_prune_label"
stageLabelName = "compaction_stage"
nodeIDLabelName = "node_id"
nodeHostLabelName = "node_host"
statusLabelName = "status"
causeLabelName = "cause"
indexTaskStatusLabelName = "index_task_status"
msgTypeLabelName = "msg_type"
collectionIDLabelName = "collection_id"
fieldIDLabelName = "field_id"
channelNameLabelName = "channel_name"
functionLabelName = "function_name"
queryTypeLabelName = "query_type"
chainLevelLabelName = "chain_level"
collectionName = "collection_name"
databaseLabelName = "db_name"
ResourceGroupLabelName = "rg"
indexName = "index_name"
isVectorIndex = "is_vector_index"
segmentStateLabelName = "segment_state"
segmentLevelLabelName = "segment_level"
segmentIsSortedLabelName = "segment_is_sorted"
segmentStorageVersionLabelName = "segment_storage_version"
segmentFormatLabelName = "segment_format"
usernameLabelName = "username"
roleNameLabelName = "role_name"
cacheNameLabelName = "cache_name"
cacheStateLabelName = "cache_state"
dataSourceLabelName = "data_source"
dataTypeLabelName = "data_type"
importStageLabelName = "import_stage"
requestScope = "scope"
fullMethodLabelName = "full_method"
reduceLevelName = "reduce_level"
reduceType = "reduce_type"
lockName = "lock_name"
lockSource = "lock_source"
lockType = "lock_type"
lockOp = "lock_op"
loadTypeName = "load_type"
pathLabelName = "path"
cgoNameLabelName = `cgo_name`
cgoTypeLabelName = `cgo_type`
queueTypeLabelName = `queue_type`
poolNameLabelName = "pool_name"
outcomeLabelName = "outcome"
// model function/UDF labels
functionTypeName = "function_type_name"
functionProvider = "function_provider"
functionName = "function_name"
// entities label
LoadedLabel = "loaded"
NumEntitiesAllLabel = "all"
TaskTypeLabel = "task_type"
TaskStateLabel = "task_state"
filesystemKeyLabelName = "fs"
reasonLabelName = "reason"
)
var (
// buckets involves durations in milliseconds,
// [1 2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 1.31072e+05]
buckets = prometheus.ExponentialBuckets(1, 2, 18)
// subMsBuckets extends buckets with sub-millisecond boundaries for histograms
// whose observations can fall below 1ms (e.g. in-memory or queue-wait paths).
// [0.1 0.25 0.5 1 2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 1.31072e+05]
subMsBuckets = append([]float64{0.1, 0.25, 0.5}, prometheus.ExponentialBuckets(1, 2, 18)...)
// longTaskBuckets provides long task duration in milliseconds
longTaskBuckets = []float64{1, 100, 500, 1000, 5000, 10000, 20000, 50000, 100000, 250000, 500000, 1000000, 3600000, 5000000, 10000000} // unit milliseconds
// size provides size in byte
sizeBuckets = []float64{10000, 100000, 1000000, 100000000, 500000000, 1024000000, 2048000000, 4096000000, 10000000000, 50000000000} // unit byte
NumNodes = prometheus.NewGaugeVec(
prometheus.GaugeOpts{
Namespace: milvusNamespace,
Name: "num_node",
Help: "number of nodes and coordinates",
}, []string{nodeIDLabelName, roleNameLabelName})
LockCosts = prometheus.NewGaugeVec(
prometheus.GaugeOpts{
Namespace: milvusNamespace,
Name: "lock_time_cost",
Help: "time cost for various kinds of locks",
}, []string{
lockName,
lockSource,
lockType,
lockOp,
})
metricRegisterer prometheus.Registerer
)
// GetRegisterer returns the global prometheus registerer
// metricsRegistry must be call after Register is called or no Register is called.
func GetRegisterer() prometheus.Registerer {
if metricRegisterer == nil {
return prometheus.DefaultRegisterer
}
return metricRegisterer
}
// Register serves prometheus http service
// Should be called by init function.
func Register(r prometheus.Registerer) {
r.MustRegister(NumNodes)
r.MustRegister(LockCosts)
r.MustRegister(BuildInfo)
r.MustRegister(RuntimeInfo)
r.MustRegister(ThreadNum)
r.MustRegister(ThreadCPUActiveNumByPool)
metricRegisterer = r
}