Skip to content

Add querier.ingester-query-max-attempts to retry on partial data. #6714

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 8 commits into
base: master
Choose a base branch
from
Open

Add querier.ingester-query-max-attempts to retry on partial data. #6714

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
aab648e
Add static retry
justinjung04 Apr 21, 2025
51aaf24
Add querier.ingester-query-max-attempts
justinjung04 Apr 22, 2025
443c6a6
Add docs
justinjung04 Apr 22, 2025
2158ca7
Add changelog
justinjung04 Apr 22, 2025
e057461
Lint
justinjung04 Apr 22, 2025
23cf005
Use backoff
justinjung04 Apr 22, 2025
761741f
Fix test
justinjung04 Apr 22, 2025
883d8bc
Change default retry from 3 to 1
justinjung04 Apr 22, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@
* [ENHANCEMENT] OTLP: Support otlp metadata ingestion. #6617
* [ENHANCEMENT] AlertManager: Add `keep_instance_in_the_ring_on_shutdown` and `tokens_file_path` configs for alertmanager ring. #6628
* [ENHANCEMENT] Querier: Add metric and enhanced logging for query partial data. #6676
* [ENHANCEMENT] Querier: Add `querier.ingester-query-max-attempts` to retry on partial data. #6714
* [BUGFIX] Ingester: Avoid error or early throttling when READONLY ingesters are present in the ring #6517
* [BUGFIX] Ingester: Fix labelset data race condition. #6573
* [BUGFIX] Compactor: Cleaner should not put deletion marker for blocks with no-compact marker. #6576
Expand Down
5 changes: 5 additions & 0 deletions docs/blocks-storage/querier.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -237,6 +237,11 @@ querier:
# CLI flag: -querier.store-gateway-consistency-check-max-attempts
[store_gateway_consistency_check_max_attempts: <int> | default = 3]

# The maximum number of times we attempt fetching data from ingesters for
# retryable errors.
# CLI flag: -querier.ingester-query-max-attempts
[ingester_query_max_attempts: <int> | default = 1]

# When distributor's sharding strategy is shuffle-sharding and this setting is
# > 0, queriers fetch in-memory series from the minimum set of required
# ingesters, selecting only ingesters which may have received series since
Expand Down
5 changes: 5 additions & 0 deletions docs/configuration/config-file-reference.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4165,6 +4165,11 @@ store_gateway_client:
# CLI flag: -querier.store-gateway-consistency-check-max-attempts
[store_gateway_consistency_check_max_attempts: <int> | default = 3]

# The maximum number of times we attempt fetching data from ingesters for
# retryable errors.
# CLI flag: -querier.ingester-query-max-attempts
[ingester_query_max_attempts: <int> | default = 1]

# When distributor's sharding strategy is shuffle-sharding and this setting is >
# 0, queriers fetch in-memory series from the minimum set of required ingesters,
# selecting only ingesters which may have received series since 'now - lookback
Expand Down
142 changes: 109 additions & 33 deletions pkg/querier/distributor_queryable.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,10 +20,14 @@ import (
"github.com/cortexproject/cortex/pkg/querier/series"
"github.com/cortexproject/cortex/pkg/tenant"
"github.com/cortexproject/cortex/pkg/util"
"github.com/cortexproject/cortex/pkg/util/backoff"
"github.com/cortexproject/cortex/pkg/util/chunkcompat"
"github.com/cortexproject/cortex/pkg/util/spanlogger"
)

const retryMinBackoff = time.Second
const retryMaxBackoff = 5 * time.Second
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ingester queries usually finish in ms. I am not sure if it is worth it to wait for 1s ~ 5s backoff retry as it may cause more issues like increased inflight queries on Ingesters.


// Distributor is the read interface to the distributor, made an interface here
// to reduce package coupling.
type Distributor interface {
Expand All @@ -38,36 +42,39 @@ type Distributor interface {
MetricsMetadata(ctx context.Context, req *client.MetricsMetadataRequest) ([]scrape.MetricMetadata, error)
}

func newDistributorQueryable(distributor Distributor, streamingMetdata bool, labelNamesWithMatchers bool, iteratorFn chunkIteratorFunc, queryIngestersWithin time.Duration, isPartialDataEnabled partialdata.IsCfgEnabledFunc) QueryableWithFilter {
func newDistributorQueryable(distributor Distributor, streamingMetdata bool, labelNamesWithMatchers bool, iteratorFn chunkIteratorFunc, queryIngestersWithin time.Duration, isPartialDataEnabled partialdata.IsCfgEnabledFunc, ingesterQueryMaxAttempts int) QueryableWithFilter {
return distributorQueryable{
distributor: distributor,
streamingMetdata: streamingMetdata,
labelNamesWithMatchers: labelNamesWithMatchers,
iteratorFn: iteratorFn,
queryIngestersWithin: queryIngestersWithin,
isPartialDataEnabled: isPartialDataEnabled,
distributor: distributor,
streamingMetdata: streamingMetdata,
labelNamesWithMatchers: labelNamesWithMatchers,
iteratorFn: iteratorFn,
queryIngestersWithin: queryIngestersWithin,
isPartialDataEnabled: isPartialDataEnabled,
ingesterQueryMaxAttempts: ingesterQueryMaxAttempts,
}
}

type distributorQueryable struct {
distributor Distributor
streamingMetdata bool
labelNamesWithMatchers bool
iteratorFn chunkIteratorFunc
queryIngestersWithin time.Duration
isPartialDataEnabled partialdata.IsCfgEnabledFunc
distributor Distributor
streamingMetdata bool
labelNamesWithMatchers bool
iteratorFn chunkIteratorFunc
queryIngestersWithin time.Duration
isPartialDataEnabled partialdata.IsCfgEnabledFunc
ingesterQueryMaxAttempts int
}

func (d distributorQueryable) Querier(mint, maxt int64) (storage.Querier, error) {
return &distributorQuerier{
distributor: d.distributor,
mint: mint,
maxt: maxt,
streamingMetadata: d.streamingMetdata,
labelNamesMatchers: d.labelNamesWithMatchers,
chunkIterFn: d.iteratorFn,
queryIngestersWithin: d.queryIngestersWithin,
isPartialDataEnabled: d.isPartialDataEnabled,
distributor: d.distributor,
mint: mint,
maxt: maxt,
streamingMetadata: d.streamingMetdata,
labelNamesMatchers: d.labelNamesWithMatchers,
chunkIterFn: d.iteratorFn,
queryIngestersWithin: d.queryIngestersWithin,
isPartialDataEnabled: d.isPartialDataEnabled,
ingesterQueryMaxAttempts: d.ingesterQueryMaxAttempts,
}, nil
}

Expand All @@ -77,13 +84,14 @@ func (d distributorQueryable) UseQueryable(now time.Time, _, queryMaxT int64) bo
}

type distributorQuerier struct {
distributor Distributor
mint, maxt int64
streamingMetadata bool
labelNamesMatchers bool
chunkIterFn chunkIteratorFunc
queryIngestersWithin time.Duration
isPartialDataEnabled partialdata.IsCfgEnabledFunc
distributor Distributor
mint, maxt int64
streamingMetadata bool
labelNamesMatchers bool
chunkIterFn chunkIteratorFunc
queryIngestersWithin time.Duration
isPartialDataEnabled partialdata.IsCfgEnabledFunc
ingesterQueryMaxAttempts int
}

// Select implements storage.Querier interface.
Expand Down Expand Up @@ -150,7 +158,9 @@ func (q *distributorQuerier) Select(ctx context.Context, sortSeries bool, sp *st
}

func (q *distributorQuerier) streamingSelect(ctx context.Context, sortSeries, partialDataEnabled bool, minT, maxT int64, matchers []*labels.Matcher) storage.SeriesSet {
results, err := q.distributor.QueryStream(ctx, model.Time(minT), model.Time(maxT), partialDataEnabled, matchers...)
results, err := q.queryWithRetry(ctx, func() (*client.QueryStreamResponse, error) {
return q.distributor.QueryStream(ctx, model.Time(minT), model.Time(maxT), partialDataEnabled, matchers...)
})

if err != nil && !partialdata.IsPartialDataError(err) {
return storage.ErrSeriesSet(err)
Expand Down Expand Up @@ -192,6 +202,33 @@ func (q *distributorQuerier) streamingSelect(ctx context.Context, sortSeries, pa
return seriesSet
}

func (q *distributorQuerier) queryWithRetry(ctx context.Context, queryFunc func() (*client.QueryStreamResponse, error)) (*client.QueryStreamResponse, error) {
if q.ingesterQueryMaxAttempts == 1 {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We need to handle case of ingesterQueryMaxAttempts set to 0 as it retries forever IIUC

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I added the validation in config.Validate:

	if cfg.IngesterQueryMaxAttempts < 1 {
		return errInvalidIngesterQueryMaxAttempts
	}

But would you still want the condition to be changed to if q.ingesterQueryMaxAttempts <= 1 {?

return queryFunc()
}

var result *client.QueryStreamResponse
var err error

retries := backoff.New(ctx, backoff.Config{
MinBackoff: retryMinBackoff,
MaxBackoff: retryMaxBackoff,
MaxRetries: q.ingesterQueryMaxAttempts,
})

for retries.Ongoing() {
result, err = queryFunc()

if err == nil || !q.isRetryableError(err) {
return result, err
}

retries.Wait()
}

return result, err
}

func (q *distributorQuerier) LabelValues(ctx context.Context, name string, hints *storage.LabelHints, matchers ...*labels.Matcher) ([]string, annotations.Annotations, error) {
var (
lvs []string
Expand All @@ -201,9 +238,13 @@ func (q *distributorQuerier) LabelValues(ctx context.Context, name string, hints
partialDataEnabled := q.partialDataEnabled(ctx)

if q.streamingMetadata {
lvs, err = q.distributor.LabelValuesForLabelNameStream(ctx, model.Time(q.mint), model.Time(q.maxt), model.LabelName(name), hints, partialDataEnabled, matchers...)
lvs, err = q.labelsWithRetry(ctx, func() ([]string, error) {
return q.distributor.LabelValuesForLabelNameStream(ctx, model.Time(q.mint), model.Time(q.maxt), model.LabelName(name), hints, partialDataEnabled, matchers...)
})
} else {
lvs, err = q.distributor.LabelValuesForLabelName(ctx, model.Time(q.mint), model.Time(q.maxt), model.LabelName(name), hints, partialDataEnabled, matchers...)
lvs, err = q.labelsWithRetry(ctx, func() ([]string, error) {
return q.distributor.LabelValuesForLabelName(ctx, model.Time(q.mint), model.Time(q.maxt), model.LabelName(name), hints, partialDataEnabled, matchers...)
})
}

if partialdata.IsPartialDataError(err) {
Expand All @@ -230,9 +271,13 @@ func (q *distributorQuerier) LabelNames(ctx context.Context, hints *storage.Labe
)

if q.streamingMetadata {
ln, err = q.distributor.LabelNamesStream(ctx, model.Time(q.mint), model.Time(q.maxt), hints, partialDataEnabled, matchers...)
ln, err = q.labelsWithRetry(ctx, func() ([]string, error) {
return q.distributor.LabelNamesStream(ctx, model.Time(q.mint), model.Time(q.maxt), hints, partialDataEnabled, matchers...)
})
} else {
ln, err = q.distributor.LabelNames(ctx, model.Time(q.mint), model.Time(q.maxt), hints, partialDataEnabled, matchers...)
ln, err = q.labelsWithRetry(ctx, func() ([]string, error) {
return q.distributor.LabelNames(ctx, model.Time(q.mint), model.Time(q.maxt), hints, partialDataEnabled, matchers...)
})
}

if partialdata.IsPartialDataError(err) {
Expand All @@ -243,6 +288,33 @@ func (q *distributorQuerier) LabelNames(ctx context.Context, hints *storage.Labe
return ln, nil, err
}

func (q *distributorQuerier) labelsWithRetry(ctx context.Context, labelsFunc func() ([]string, error)) ([]string, error) {
if q.ingesterQueryMaxAttempts == 1 {
return labelsFunc()
}

var result []string
var err error

retries := backoff.New(ctx, backoff.Config{
MinBackoff: retryMinBackoff,
MaxBackoff: retryMaxBackoff,
MaxRetries: q.ingesterQueryMaxAttempts,
})

for retries.Ongoing() {
result, err = labelsFunc()

if err == nil || !q.isRetryableError(err) {
return result, err
}

retries.Wait()
}

return result, err
}

// labelNamesWithMatchers performs the LabelNames call by calling ingester's MetricsForLabelMatchers method
func (q *distributorQuerier) labelNamesWithMatchers(ctx context.Context, hints *storage.LabelHints, partialDataEnabled bool, matchers ...*labels.Matcher) ([]string, annotations.Annotations, error) {
log, ctx := spanlogger.New(ctx, "distributorQuerier.labelNamesWithMatchers")
Expand Down Expand Up @@ -297,6 +369,10 @@ func (q *distributorQuerier) partialDataEnabled(ctx context.Context) bool {
return q.isPartialDataEnabled != nil && q.isPartialDataEnabled(userID)
}

func (q *distributorQuerier) isRetryableError(err error) bool {
return partialdata.IsPartialDataError(err)
}

type distributorExemplarQueryable struct {
distributor Distributor
}
Expand Down
Loading
Loading