gsadmin/rustfs
1
0
Fork 0
mirror of https://github.com/rustfs/rustfs.git synced 2026-08-31 09:18:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: decouple readiness from cluster health (#3912)

Browse Source 
* fix: include foreground write pressure

* fix: split readiness and cluster health probes

* fix: publish ready on node readiness

* fix: bound cluster health collection

* test: pin health probe collector routing

* fix: qualify app storage ECStore type

* test(admin): narrow runtime capabilities topology assertion

---------

Co-authored-by: houseme <housemecn@gmail.com>
This commit is contained in:
cxymds
2026-06-26 21:55:18 +08:00
committed by GitHub
parent 6469d6ada8
commit 7820a55fdc
14 changed files with 769 additions and 165 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/config/src/constants/heal.rs
+7
View File
@@ -140,6 +140,10 @@ pub const ENV_HEAL_MAINLINE_THROTTLE_ENABLE: &str = "RUSTFS_HEAL_MAINLINE_THROTT
/// at which background heal work pauses starting new tasks.
pub const ENV_HEAL_MAINLINE_READ_UTILIZATION_HIGH_PERCENT: &str = "RUSTFS_HEAL_MAINLINE_READ_UTILIZATION_HIGH_PERCENT";
/// Environment variable that controls the foreground write utilization percentage
/// at which background heal work pauses starting new tasks.
pub const ENV_HEAL_MAINLINE_WRITE_UTILIZATION_HIGH_PERCENT: &str = "RUSTFS_HEAL_MAINLINE_WRITE_UTILIZATION_HIGH_PERCENT";
/// Environment variable that controls how soon the heal scheduler rechecks foreground
/// pressure after delaying background work.
pub const ENV_HEAL_MAINLINE_MAX_SLEEP_MS: &str = "RUSTFS_HEAL_MAINLINE_MAX_SLEEP_MS";
@@ -168,5 +172,8 @@ pub const DEFAULT_HEAL_MAINLINE_THROTTLE_ENABLE: bool = true;
/// Default foreground read permit utilization threshold for pausing best-effort heal task starts.
pub const DEFAULT_HEAL_MAINLINE_READ_UTILIZATION_HIGH_PERCENT: usize = 80;
/// Default foreground write utilization threshold for pausing best-effort heal task starts.
pub const DEFAULT_HEAL_MAINLINE_WRITE_UTILIZATION_HIGH_PERCENT: usize = 80;
/// Default foreground pressure recheck delay for heal scheduler, in milliseconds.
pub const DEFAULT_HEAL_MAINLINE_MAX_SLEEP_MS: u64 = 250;
crates/config/src/constants/health.rs
+5
View File
@@ -22,6 +22,11 @@ pub const DEFAULT_HEALTH_ENDPOINT_ENABLE: bool = true;
pub const ENV_HEALTH_READINESS_CACHE_TTL_MS: &str = "RUSTFS_HEALTH_READINESS_CACHE_TTL_MS";
pub const DEFAULT_HEALTH_READINESS_CACHE_TTL_MS: u64 = 1000;
/// Timeout for cluster health readiness collectors (milliseconds).
/// This bounds expensive storage and lock quorum checks used by cluster probes.
pub const ENV_HEALTH_CLUSTER_TIMEOUT_MS: &str = "RUSTFS_HEALTH_CLUSTER_TIMEOUT_MS";
pub const DEFAULT_HEALTH_CLUSTER_TIMEOUT_MS: u64 = 2000;
/// Enable minimal health payload mode for GET `/health*` responses.
/// When enabled, only `status` and `ready` fields are returned.
pub const ENV_HEALTH_MINIMAL_RESPONSE_ENABLE: &str = "RUSTFS_HEALTH_MINIMAL_RESPONSE_ENABLE";