feat: add configurable ParquetMergePolicyConfig to index settings

Adds `parquet_merge_policy` section to `IndexingSettings`, making the
Parquet merge policy configurable per-index via YAML. Parameters:

- merge_factor (default 10): min splits to trigger a merge
- max_merge_factor (default 12): max splits per merge
- max_merge_ops (default 4): bounds write amplification
- target_split_size_bytes (default 256 MiB): target output size
- maturation_period (default 48h): split maturity timeout
- max_finalize_merge_operations (default 3): cold-window shutdown limit

Mirrors the existing merge_policy config pattern for logs/traces.
Updates index-config.md documentation with the new section.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: g-talbot

docs/configuration/index-config.md

@@ -594,7 +594,8 @@ This section describes indexing settings for a given index.
 | ------------- | ------------- | ------------- |
 | `commit_timeout_secs`      | Maximum number of seconds before committing a split since its creation.   | `60` |
 | `split_num_docs_target` | Target number of docs per split.   | `10000000` |
-| `merge_policy` | Describes the strategy used to trigger split merge operations (see [Merge policies](#merge-policies) section below). |
+| `merge_policy` | Describes the strategy used to trigger split merge operations for logs/traces (see [Merge policies](#merge-policies) section below). |
+| `parquet_merge_policy` | Describes the merge policy for Parquet (metrics/sketches) splits (see [Parquet merge policy](#parquet-merge-policy) section below). |
 | `resources.heap_size`      | Indexer heap size per source per index.   | `2000000000` |
 | `docstore_compression_level` | Level of compression used by zstd for the docstore. Lower values may increase ingest speed, at the cost of index size | `8` |
 | `docstore_blocksize` | Size of blocks in the docstore, in bytes. Lower values may improve doc retrieval speed, at the cost of index size | `1000000` |
@@ -687,6 +688,35 @@ indexing_settings:
         type: "no_merge"
 ```
 
+#### Parquet merge policy
+
+*For indexes using the Parquet indexing pipeline (metrics, sketches).*
+
+The Parquet merge policy controls how Parquet splits within a compaction scope (same time window, partition, and sort schema) are merged. It uses a constant write amplification strategy: splits at the same merge level are greedily accumulated until reaching `max_merge_factor` or `target_split_size_bytes`.
+
+```yaml
+version: 0.7
+index_id: "my-metrics-index"
+# ...
+indexing_settings:
+  parquet_merge_policy:
+    merge_factor: 10
+    max_merge_factor: 12
+    max_merge_ops: 4
+    target_split_size_bytes: 268435456
+    maturation_period: 48h
+    max_finalize_merge_operations: 3
+```
+
+
+| Variable      | Description   | Default value |
+| ------------- | ------------- | ------------- |
+| `merge_factor` | Minimum number of splits to trigger a merge. | `10` |
+| `max_merge_factor` | Maximum number of splits in a single merge operation. | `12` |
+| `max_merge_ops` | Maximum number of merges a split can undergo before becoming mature. Bounds total write amplification. | `4` |
+| `target_split_size_bytes` | Target size for merged output splits in bytes. Merges trigger when accumulated bytes reach this threshold, even if `merge_factor` is not reached. | `268435456` (256 MiB) |
+| `maturation_period` | Duration after creation when a split becomes mature (never merged again). | `48h` |
+| `max_finalize_merge_operations` | *(advanced)* Maximum number of merge operations emitted during cold-window finalization at pipeline shutdown. Set to `0` to disable. | `3` |
 
 
 ### Indexer memory usage

quickwit/quickwit-config/src/index_config/mod.rs

@@ -36,7 +36,7 @@ use siphasher::sip::SipHasher;
 use tracing::warn;
 
 use crate::index_config::serialize::VersionedIndexConfig;
-use crate::merge_policy_config::MergePolicyConfig;
+use crate::merge_policy_config::{MergePolicyConfig, ParquetMergePolicyConfig};
 
 #[derive(Clone, Debug, Serialize, Deserialize, utoipa::ToSchema)]
 #[serde(deny_unknown_fields)]
@@ -118,6 +118,11 @@ pub struct IndexingSettings {
     pub split_num_docs_target: usize,
     #[serde(default)]
     pub merge_policy: MergePolicyConfig,
+    /// Merge policy for Parquet (metrics/sketches) splits. Controls how
+    /// Parquet splits are compacted within time windows. Only used by
+    /// indexes that use the Parquet indexing pipeline.
+    #[serde(default)]
+    pub parquet_merge_policy: ParquetMergePolicyConfig,
     #[serde(default)]
     pub resources: IndexingResources,
 }
@@ -160,6 +165,7 @@ impl Default for IndexingSettings {
             docstore_compression_level: Self::default_docstore_compression_level(),
             split_num_docs_target: Self::default_split_num_docs_target(),
             merge_policy: MergePolicyConfig::default(),
+            parquet_merge_policy: ParquetMergePolicyConfig::default(),
             resources: IndexingResources::default(),
         }
     }

quickwit/quickwit-config/src/lib.rs

@@ -67,7 +67,8 @@ use tracing::warn;
 use crate::index_template::IndexTemplateV0_8;
 pub use crate::index_template::{IndexTemplate, IndexTemplateId, VersionedIndexTemplate};
 use crate::merge_policy_config::{
-    ConstWriteAmplificationMergePolicyConfig, MergePolicyConfig, StableLogMergePolicyConfig,
+    ConstWriteAmplificationMergePolicyConfig, MergePolicyConfig, ParquetMergePolicyConfig,
+    StableLogMergePolicyConfig,
 };
 pub use crate::metastore_config::{
     MetastoreBackend, MetastoreConfig, MetastoreConfigs, PostgresMetastoreConfig,
@@ -113,6 +114,7 @@ pub fn disable_ingest_v1() -> bool {
     KafkaSourceParams,
     KinesisSourceParams,
     MergePolicyConfig,
+    ParquetMergePolicyConfig,
     PubSubSourceParams,
     PulsarSourceAuth,
     PulsarSourceParams,

quickwit/quickwit-config/src/merge_policy_config.rs

@@ -119,6 +119,70 @@ impl Default for StableLogMergePolicyConfig {
     }
 }
 
+// --- Parquet merge policy config ---
+
+fn default_target_split_size_bytes() -> u64 {
+    256 * 1024 * 1024 // 256 MiB
+}
+
+fn default_max_finalize_merge_operations() -> usize {
+    3
+}
+
+/// Configuration for the Parquet (metrics/sketches) merge policy.
+///
+/// Controls how Parquet splits within a compaction scope are merged.
+/// Splits at the same `num_merge_ops` level are greedily accumulated
+/// until reaching `max_merge_factor` or `target_split_size_bytes`.
+#[derive(Debug, Clone, Serialize, Deserialize, Eq, PartialEq, Hash, utoipa::ToSchema)]
+#[serde(deny_unknown_fields)]
+pub struct ParquetMergePolicyConfig {
+    /// Minimum number of splits to trigger a merge.
+    #[serde(default = "default_merge_factor")]
+    pub merge_factor: usize,
+    /// Maximum number of splits in a single merge operation.
+    #[serde(default = "default_max_merge_factor")]
+    pub max_merge_factor: usize,
+    /// Maximum number of merges a split can undergo before becoming mature.
+    /// Bounds total write amplification.
+    #[serde(default = "default_parquet_max_merge_ops")]
+    pub max_merge_ops: u32,
+    /// Target size for merged output splits in bytes. Merges are triggered
+    /// when accumulated bytes reach this threshold, even if `merge_factor`
+    /// is not reached.
+    #[serde(default = "default_target_split_size_bytes")]
+    pub target_split_size_bytes: u64,
+    /// Duration after creation when a split becomes mature regardless of
+    /// size or merge count. Mature splits are never merged.
+    #[schema(value_type = String)]
+    #[serde(default = "default_maturation_period")]
+    #[serde(deserialize_with = "parse_human_duration")]
+    #[serde(serialize_with = "serialize_duration")]
+    pub maturation_period: Duration,
+    /// Maximum number of merge operations emitted during cold-window
+    /// finalization at shutdown. Set to 0 to disable.
+    #[serde(default = "default_max_finalize_merge_operations")]
+    #[serde(skip_serializing_if = "is_zero")]
+    pub max_finalize_merge_operations: usize,
+}
+
+fn default_parquet_max_merge_ops() -> u32 {
+    4
+}
+
+impl Default for ParquetMergePolicyConfig {
+    fn default() -> Self {
+        Self {
+            merge_factor: default_merge_factor(),
+            max_merge_factor: default_max_merge_factor(),
+            max_merge_ops: default_parquet_max_merge_ops(),
+            target_split_size_bytes: default_target_split_size_bytes(),
+            maturation_period: default_maturation_period(),
+            max_finalize_merge_operations: default_max_finalize_merge_operations(),
+        }
+    }
+}
+
 fn parse_human_duration<'de, D>(deserializer: D) -> Result<Duration, D::Error>
 where D: Deserializer<'de> {
     let value: String = Deserialize::deserialize(deserializer)?;