perf(delivery): allocate node CPU by maintenance phase
This commit is contained in:
@@ -49,10 +49,11 @@ START_ORDER = (
|
||||
"worker",
|
||||
"proxy",
|
||||
)
|
||||
# Root has no DB/MQ dependency. Pair ordinary full-bundle startup checks under
|
||||
# the two-CPU node cap; every unit keeps its original startup deadline.
|
||||
# PG and MQ each allow one CPU during full-bundle startup checks. Start Root
|
||||
# after them, then pair ordinary checks within the unchanged two-CPU node cap.
|
||||
START_GROUPS = (
|
||||
(*DEPENDENCIES, "sandbox-factory"),
|
||||
DEPENDENCIES,
|
||||
("sandbox-factory",),
|
||||
("coordinator", "api"),
|
||||
("outbox", "problem-import"),
|
||||
("worker",),
|
||||
|
||||
Reference in New Issue
Block a user