admin/openhuman
1
0
Fork 0
mirror of https://github.com/tinyhumansai/openhuman.git synced 2026-07-27 21:08:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(memory): route only unrecoverable job failures to error status (#3365) (#3630)

Browse Source 
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Steven Enamakel <enamakel@tinyhumans.ai>
This commit is contained in:
sanil-23
2026-06-13 13:28:15 -07:00
committed by GitHub
co-authored by Claude Steven Enamakel
parent e105e63c0c
commit 595fb41c9b
3 changed files with 172 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/memory_tree_sync_deep_raw_coverage_e2e.rs
+24 -2
View File
@@ -499,6 +499,8 @@ async fn memory_tree_rpc_status_set_enabled_backfill_and_ingest_errors() {
.expect("build failed job"),
)
.expect("enqueue failed job");
// #3365: an untyped/transient failed job (no `failure_class`) self-heals via
// requeue, so it surfaces as `degraded` ("retrying"), not a hard `error`.
with_connection(&cfg, |conn| {
conn.execute(
"UPDATE mem_tree_jobs
@@ -509,10 +511,30 @@ async fn memory_tree_rpc_status_set_enabled_backfill_and_ingest_errors() {
)?;
Ok(())
})
.expect("mark failed");
.expect("mark failed (untyped)");
let degraded = pipeline_status_rpc(&cfg)
.await
.expect("degraded status")
.value;
assert_eq!(degraded.status, "degraded");
assert!(degraded.reason.unwrap().contains("retrying"));
// #3365: an UNRECOVERABLE failed job (budget / auth / dim-mismatch) stays
// parked and is user-actionable, so it escalates to `error`.
with_connection(&cfg, |conn| {
conn.execute(
"UPDATE mem_tree_jobs
SET failure_class = 'unrecoverable'
WHERE kind = 'extract_chunk'
AND payload_json LIKE '%round18-failed%'",
[],
)?;
Ok(())
})
.expect("mark unrecoverable");
let errored = pipeline_status_rpc(&cfg).await.expect("error status").value;
assert_eq!(errored.status, "error");
assert!(errored.reason.unwrap().contains("failed job"));
assert!(errored.reason.unwrap().contains("unrecoverable"));
cfg.scheduler_gate.mode = SchedulerGateMode::Off;
let paused = pipeline_status_rpc(&cfg)