You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: web/apps/docs/content/docs/reference/nativelink-config/main.mdx
+3-1Lines changed: 3 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -5,7 +5,7 @@ full: true
5
5
---
6
6
7
7
{/* AUTOGENERATED — do not edit by hand.
8
-
Source: nativelink-config @ main (64ca9740)
8
+
Source: nativelink-config @ main (712f3215)
9
9
Regenerate from web/: bun --filter @nativelink/docs gen:config-reference */}
10
10
11
11
<ConfigVersionSwitcher />
@@ -878,6 +878,7 @@ See [`experimental_cloud_object_store`](#experimental_cloud_object_store-1) for
878
878
|`health_check_timeout_ms`| integer (uint64) | — | 4000 (4 seconds) | Per-call ceiling for the `check_health` PING in milliseconds. |
879
879
|`read_chunk_size`| integer (uint) | — | 64KiB | The amount of data to read from the Redis server at a time. This is used to limit the amount of memory used when reading large objects from the Redis server as well as limiting the amount of time a single read operation can take. |
880
880
|`connection_pool_size`| integer (uint) | — | 3 | The number of connections to keep open to the Redis servers. |
881
+
|`key_ttl_s`| integer (uint64) | — | 0 (keys never expire) | Expire keys written by this store after this many seconds. |
881
882
|`max_chunk_uploads_per_update`| integer (uint) | — | 10 | The maximum number of upload chunks to allow per update. This is used to limit the amount of memory used when uploading large objects to the Redis server. A good rule of thumb is to think of the data as: `AVAIL_MEMORY / (read_chunk_size * max_chunk_uploads_per_update) = THORETICAL_MAX_CONCURRENT_UPLOADS` (note: it's a good idea to divide `AVAIL_MAX_MEMORY` by ~10 to account for other memory usage) |
882
883
|`scan_count`| integer (uint) | — | 10000 | The COUNT value passed when scanning keys in Redis. This is used to hint the amount of work that should be done per response. |
883
884
|`retry`|[Retry](#retry)| — | — | Retry configuration to use when a network request fails. |
@@ -929,6 +930,7 @@ Configuration for `ExperimentalMongoDB` store.
929
930
|`platform_properties`| map of string to [WorkerProperty](#workerproperty)| Yes | — | Properties of this worker. This configuration will be sent to the scheduler and used to tell the scheduler to restrict what should be executed on this worker. |
930
931
|`additional_environment`| map of string to [EnvironmentSource](#environmentsource)| — | — | An optional mapping of environment names to set for the execution as well as those specified in the action itself. If set, will set each key as an environment variable before executing the job with the value of the environment variable being the value of the property of the action being executed of that name or the fixed value. |
931
932
|`directory_cache`|[DirectoryCacheConfig](#directorycacheconfig)| — | — | Optional directory cache configuration for improving performance by caching reconstructed input directories and using hardlinks instead of rebuilding them from CAS for every action. |
933
+
|`experimental_active_input_leases`| boolean | — | false (eviction behavior is unchanged) | Optional and experimental: lease every digest in an active action's input Merkle closure in the worker's locally eviction-managed CAS tiers (the `cas_fast_slow_store`'s filesystem-backed stores) until the action has finished cleanup. This prevents `Lost inputs no longer available remotely` failures caused by local fast-tier eviction while inputs are being materialized under cache pressure. |
932
934
|`use_namespaces`| boolean | — | False | Whether to use namespaces to isolate the execution. This is only available on Linux. It is highly recommended as it avoids a number of issues with zombie processes and also provides additional hermeticity. If explicitly set to true and it is not supported the worker will exit with an error. |
933
935
|`use_mount_namespace`| boolean | — | False | Whether to use a mount namespace to isolate the worker root. This is only available on Linux and when `use_namespaces` is true. It is highly recommended provides additional hermeticity. If explicitly set to true and it is not supported or `use_namespaces` is not set to true the worker will exit with an error. |
0 commit comments