Commit 3324da1
KP13 listed this on #366 as "Lower SERVICE_WAIT from 30→10s now that
'sun' is gone (the only remaining required service, input_boolean,
resolves in 1s in every observed run)":
#366 (comment)
Symbol-name note: KP13's "SERVICE_WAIT" is the colloquial label; the
live constant in tests/src/e2e/conftest.py:1104 is INPUT_BOOLEAN_WAIT
(introduced after #1227 dropped "sun" from REQUIRED_SERVICES, leaving
input_boolean as the only service-domain wait).
10s preserves a 10× safety margin over the 1s observed-resolve KP13
cited. The timeout branch is non-fatal (logger.warning +
_dump_ha_readiness_diagnostics, no pytest.fail at this gate), so a
tighter budget cannot regress the suite: at worst a slow-boot run
logs an earlier warning while downstream tests continue, and the
HA_MCP_TOOLS_WAIT gate further down keeps its full 180s pytest.fail-
backed budget for the surface that does block the run.
Sibling 30s constants in the same fixture are intentionally not
touched:
* STABILIZATION_TIMEOUT (line 1007)
* ENTITY_STABILIZATION_TIMEOUT (line 1057)
* SUN_WAIT (line 1181) — state-readiness wait for the sun.sun
entity, not a service-domain wait (different concern from the
one KP13 scoped to SERVICE_WAIT)
KP13's same comment also requests a broader "70+s session-init
readiness gates" audit covering STABILIZATION_TIMEOUT,
ENTITY_STABILIZATION_TIMEOUT, MIN_COMPONENTS, MIN_ENTITIES — that
audit is a separate item on his list and stays out of scope here to
keep the change reviewable as a single-constant tightening.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 11563cb commit 3324da1
1 file changed
Lines changed: 1 addition & 1 deletion
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1101 | 1101 | | |
1102 | 1102 | | |
1103 | 1103 | | |
1104 | | - | |
| 1104 | + | |
1105 | 1105 | | |
1106 | 1106 | | |
1107 | 1107 | | |
| |||
0 commit comments