feat(scheduling): replica spread + resource-aware auto placement (#56)

services.spread distributes replicas across every online, schedulable
agent (replica i → agents[i mod n], deterministic). auto placement now
orders by real headroom — memory utilisation, cpu usage, then container
count — instead of count alone.

Lifecycle (start/stop/restart/delete) and GET /runtime are now driven
by container_instances inventory, so remote replicas are visible and
tear down wherever they run — covers spread, pins, and stale pin
leftovers. Spread and explicit pins are mutually exclusive.

Generated with [Devin](https://devin.ai)

Co-authored-by: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
This commit is contained in:
Tomáš DvořákandDevin authored and GitHub committed 2026-10-09 17:03:26 +02:00
1 parent bc93c83cf9
commit dd5ce80ce2
21 files changed
+469 -100

No files matched your search

@@ -94,6 +94,8 @@ type Querier interface {
// display, when the caller owns or is a member of it, or when the caller is a
// platform admin. Anonymous callers pass user_id = uuid.Nil, is_admin = false.
ListProjectsWithStatsByUser(ctx context.Context, arg ListProjectsWithStatsByUserParams) ([]ListProjectsWithStatsByUserRow, error)
ListSchedulableAgents(ctx context.Context) ([]ListSchedulableAgentsRow, error)
ListServiceContainers(ctx context.Context, serviceID string) ([]ListServiceContainersRow, error)
ListServiceTemplatesByCategoryForUser(ctx context.Context, arg ListServiceTemplatesByCategoryForUserParams) ([]ServiceTemplate, error)
ListServiceTemplatesForUser(ctx context.Context, ownerID uuid.NullUUID) ([]ServiceTemplate, error)
ListUserInvites(ctx context.Context) ([]UserInvite, error)
@@ -102,6 +104,8 @@ type Querier interface {
MarkAllNotificationsReadByUser(ctx context.Context, arg MarkAllNotificationsReadByUserParams) error
MarkNotificationReadByIDAndUser(ctx context.Context, arg MarkNotificationReadByIDAndUserParams) error
MarkUserInviteUsed(ctx context.Context, arg MarkUserInviteUsedParams) error
// Resource-aware: lowest memory utilisation first, then cpu, then container
// count. Agents without telemetry sort last (NULL → worst score).
PickLeastLoadedAgent(ctx context.Context) (string, error)
RevokeAgentAuthToken(ctx context.Context, id uuid.UUID) (AgentAuthToken, error)
RevokeUserToken(ctx context.Context, arg RevokeUserTokenParams) (int64, error)