anti/DECNET
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

feat(fleet): auto-swarm deploy — shard across enrolled workers when master

Browse Source 
POST /deckies/deploy now branches on DECNET_MODE + enrolled host presence:
when the caller is a master with at least one reachable swarm host, round-
robin host_uuids are assigned over new deckies and the config is dispatched
via AgentClient. Falls back to local docker-compose otherwise.

Extracts the dispatch loop from api_deploy_swarm into dispatch_decnet_config
so both endpoints share the same shard/dispatch/persist path. Adds
GET /system/deployment-mode for the UI to show 'will shard across N hosts'
vs 'will deploy locally' before the operator clicks deploy.
This commit is contained in:
anti
2026-04-19 06:09:08 -04:00
parent cb1a1d1270
commit 79db999030
7 changed files with 245 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
decnet/web/router/swarm/api_deploy_swarm.py
View File

@@ -47,15 +47,18 @@ def _worker_config(base: DecnetConfig, shard: list[DeckyConfig]) -> DecnetConfig
return base.model_copy(update={"deckies": shard})
@router.post("/deploy", response_model=SwarmDeployResponse, tags=["Swarm Deployments"])
async def api_deploy_swarm(
req: SwarmDeployRequest,
repo: BaseRepository = Depends(get_repo),
async def dispatch_decnet_config(
config: DecnetConfig,
repo: BaseRepository,
dry_run: bool = False,
no_cache: bool = False,
) -> SwarmDeployResponse:
if req.config.mode != "swarm":
raise HTTPException(status_code=400, detail="mode must be 'swarm'")
"""Shard ``config`` by ``host_uuid`` and dispatch to each worker in parallel.
buckets = _shard_by_host(req.config)
Shared between POST /swarm/deploy (explicit swarm call) and the auto-swarm
branch of POST /deckies/deploy.
"""
buckets = _shard_by_host(config)
hosts: dict[str, dict[str, Any]] = {}
for host_uuid in buckets:
@@ -66,17 +69,17 @@ async def api_deploy_swarm(
async def _dispatch(host_uuid: str, shard: list[DeckyConfig]) -> SwarmHostResult:
host = hosts[host_uuid]
cfg = _worker_config(req.config, shard)
cfg = _worker_config(config, shard)
try:
async with AgentClient(host=host) as agent:
body = await agent.deploy(cfg, dry_run=req.dry_run, no_cache=req.no_cache)
body = await agent.deploy(cfg, dry_run=dry_run, no_cache=no_cache)
for d in shard:
await repo.upsert_decky_shard(
{
"decky_name": d.name,
"host_uuid": host_uuid,
"services": json.dumps(d.services),
"state": "running" if not req.dry_run else "pending",
"state": "running" if not dry_run else "pending",
"last_error": None,
"updated_at": datetime.now(timezone.utc),
}
@@ -102,3 +105,15 @@ async def api_deploy_swarm(
*(_dispatch(uuid_, shard) for uuid_, shard in buckets.items())
)
return SwarmDeployResponse(results=list(results))
@router.post("/deploy", response_model=SwarmDeployResponse, tags=["Swarm Deployments"])
async def api_deploy_swarm(
req: SwarmDeployRequest,
repo: BaseRepository = Depends(get_repo),
) -> SwarmDeployResponse:
if req.config.mode != "swarm":
raise HTTPException(status_code=400, detail="mode must be 'swarm'")
return await dispatch_decnet_config(
req.config, repo, dry_run=req.dry_run, no_cache=req.no_cache
)