[0.37-nov 19] (#106)

Author: jnyi

.mdox.validate.yaml

@@ -47,3 +47,6 @@ validators:
   # Expired certificate
   - regex: 'bestpractices\.coreinfrastructure\.org\/projects\/3048'
     type: 'ignore'
+  # Frequent DNS issues. 
+  - regex: 'build\.thebeat\.co'
+    type: 'ignore'

CHANGELOG.md

@@ -12,6 +12,16 @@ We use *breaking :warning:* to mark changes that are not backward compatible (re
 
 ### Fixed
 
+### Added
+
+### Changed
+
+### Removed
+
+## [v0.37.0 - <in progress>](https://github.com/thanos-io/thanos/tree/release-0.37)
+
+### Fixed
+
 - [#7511](https://github.com/thanos-io/thanos/pull/7511) Query Frontend: fix doubled gzip compression for response body.
 - [#7592](https://github.com/thanos-io/thanos/pull/7592) Ruler: Only increment `thanos_rule_evaluation_with_warnings_total` metric for non PromQL warnings.
 - [#7614](https://github.com/thanos-io/thanos/pull/7614) *: fix debug log formatting.
@@ -26,6 +36,8 @@ We use *breaking :warning:* to mark changes that are not backward compatible (re
 - [#7852](https://github.com/thanos-io/thanos/pull/7852) Query Frontend: pass "stats" parameter forward to queriers and fix Prometheus stats merging.
 - [#7832](https://github.com/thanos-io/thanos/pull/7832) Query Frontend: Fix cache keys for dynamic split intervals.
 - [#7885](https://github.com/thanos-io/thanos/pull/7885) Store: Return chunks to the pool after completing a Series call.
+- [#7893](https://github.com/thanos-io/thanos/pull/7893) Sidecar: Fix retrieval of external labels for Prometheus v3.0.0.
+- [#7903](https://github.com/thanos-io/thanos/pull/7903) Query: Fix panic on regex store matchers.
 
 ### Added
 - [#7763](https://github.com/thanos-io/thanos/pull/7763) Ruler: use native histograms for client latency metrics.

VERSION

@@ -1 +1 @@
-0.37.0-dev
+0.38.0-dev

cmd/thanos/main_test.go

@@ -105,6 +105,16 @@ func (b *erroringBucket) Name() string {
 	return b.bkt.Name()
 }
 
+// IterWithAttributes allows to iterate over objects in the bucket with their attributes.
+func (b *erroringBucket) IterWithAttributes(ctx context.Context, dir string, f func(objstore.IterObjectAttributes) error, options ...objstore.IterOption) error {
+	return b.bkt.IterWithAttributes(ctx, dir, f, options...)
+}
+
+// SupportedIterOptions returns the supported iteration options.
+func (b *erroringBucket) SupportedIterOptions() []objstore.IterOptionType {
+	return b.bkt.SupportedIterOptions()
+}
+
 // Ensures that downsampleBucket() stops its work properly
 // after an error occurs with some blocks in the backlog.
 // Testing for https://github.com/thanos-io/thanos/issues/4960.

cmd/thanos/store.go

@@ -21,6 +21,7 @@ import (
 	"github.com/prometheus/client_golang/prometheus"
 	commonmodel "github.com/prometheus/common/model"
 	"github.com/prometheus/common/route"
+	"gopkg.in/yaml.v2"
 
 	"github.com/thanos-io/objstore"
 	"github.com/thanos-io/objstore/client"
@@ -32,6 +33,7 @@ import (
 	"github.com/thanos-io/thanos/pkg/block/metadata"
 	"github.com/thanos-io/thanos/pkg/component"
 	hidden "github.com/thanos-io/thanos/pkg/extflag"
+	"github.com/thanos-io/thanos/pkg/exthttp"
 	"github.com/thanos-io/thanos/pkg/extkingpin"
 	"github.com/thanos-io/thanos/pkg/extprom"
 	extpromhttp "github.com/thanos-io/thanos/pkg/extprom/http"
@@ -308,8 +310,11 @@ func runStore(
 	if err != nil {
 		return err
 	}
-
-	bkt, err := client.NewBucket(logger, confContentYaml, conf.component.String(), nil)
+	customBktConfig := exthttp.DefaultCustomBucketConfig()
+	if err := yaml.Unmarshal(confContentYaml, &customBktConfig); err != nil {
+		return errors.Wrap(err, "parsing config YAML file")
+	}
+	bkt, err := client.NewBucket(logger, confContentYaml, conf.component.String(), exthttp.CreateHedgedTransportWithConfig(customBktConfig))
 	if err != nil {
 		return err
 	}

docs/components/receive.md

@@ -113,6 +113,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 

docs/components/sidecar.md

@@ -75,6 +75,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 

docs/components/store.md

@@ -34,6 +34,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 
@@ -576,6 +577,33 @@ Note that there must be no trailing slash in the `peers` configuration i.e. one
 
 If timeout is set to zero then there is no timeout for fetching and fetching's lifetime is equal to the lifetime to the original request's lifetime. It is recommended to keep it higher than zero. It is generally preferred to keep this value higher because the fetching operation potentially includes loading of data from remote object storage.
 
+## Hedged Requests
+
+Thanos Store Gateway supports `hedged requests` to enhance performance and reliability, particularly in high-latency environments. This feature addresses `long-tail latency issues` that can occur between the Thanos Store Gateway and an external cache, reducing the impact of slower response times on overall performance.
+
+The configuration options for hedged requests allow for tuning based on latency tolerance and cost considerations, as some providers may charge per request.
+
+In the `bucket.yml` file, you can specify the following fields under `hedging_config`:
+
+- `enabled`: bool to enable hedged requests.
+- `up_to`: maximum number of hedged requests allowed for each initial request.
+  - **Purpose**: controls the redundancy level of hedged requests to improve response times.
+  - **Cost vs. Benefit**: increasing up_to can reduce latency but may increase costs, as some providers charge per request. Higher values provide diminishing returns on latency beyond a certain level.
+- `quantile`: latency threshold, specified as a quantile (e.g., percentile), which determines when additional hedged requests should be sent.
+  - **Purpose**: controls when hedged requests are triggered based on response time distribution.
+  - **Cost vs. Benefit**: lower quantile (e.g., 0.7) initiates hedged requests sooner, potentially raising costs while lowering latency variance. A higher quantile (e.g., 0.95) will initiate hedged requests later, reducing cost by limiting redundancy.
+
+By default, `hedging_config` is set as follows:
+
+```yaml
+hedging_config:
+  enabled: false
+  up_to: 3
+  quantile: 0.9
+```
+
+This configuration sends up to three additional requests if the initial request response time exceeds the 90th percentile.
+
 ## Index Header
 
 In order to query series inside blocks from object storage, Store Gateway has to know certain initial info from each block index. In order to achieve so, on startup the Gateway builds an `index-header` for each block and stores it on local disk; such `index-header` is build by downloading specific pieces of original block's index, stored on local disk and then mmaped and used by Store Gateway.

docs/components/tools.md

@@ -129,6 +129,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 
@@ -700,6 +701,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 
@@ -803,6 +805,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 

docs/storage.md

@@ -103,6 +103,7 @@ config:
     kms_encryption_context: {}
     encryption_key: ""
   sts_endpoint: ""
+  max_retries: 0
 prefix: ""
 ```
 
@@ -305,6 +306,7 @@ config:
       insecure_skip_verify: false
     disable_compression: false
   chunk_size_bytes: 0
+  max_retries: 0
 prefix: ""
 ```
 
@@ -495,6 +497,7 @@ config:
   endpoint: ""
   secret_key: ""
   secret_id: ""
+  max_retries: 0
   http_config:
     idle_conn_timeout: 1m30s
     response_header_timeout: 2m