minor model updates and configuration change since timeout was added

src/fastcache_api/models.py

@@ -1,11 +1,12 @@
+import json
 from enum import StrEnum
 from pathlib import Path
 from uuid import UUID
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import AnyUrl, BaseModel, ConfigDict
 
 
-class CacheStatus(StrEnum):
+class CacheState(StrEnum):
     new = "new"
     queued = "queued"
     active = "active"
@@ -21,42 +22,30 @@ class CacheConfig(BaseModel):
     """Configuration document persisted as JSON on a cache row.
 
     Mirrors the cacheserver config (see config/default.json) plus the
-    resolved ZMQ bind URIs.
+    resolved ZMQ bind URIs. ``to_fastcache_json`` renders the on-disk config
+    the fastcache binary consumes (see src/config.cpp): the ZMQ URIs are
+    renamed to ``inurl``/``outurl`` and ``hostname`` is dropped.
     """
 
     hostname: str
-    pull_uri: str
-    push_uri: str
+    pull_uri: AnyUrl
+    push_uri: AnyUrl
     type: int = 4
     helper_threads: int = 0
-    num_workers: int = 1
     io_threads: int = 16
     hwm: int = 10
+    # Default to 2 minutes for now.
+    timeout: int = 120_000
     verbose: bool = False
 
-
-class FastcacheConfig(BaseModel):
-    """On-disk config consumed by the fastcache binary (see src/config.cpp).
-
-    We should adjust these fields later to better match the request schema.
-    """
-
-    inurl: str
-    outurl: str
-    type: int = 4
-    helper_threads: int = 0
-    num_workers: int = 1
-    io_threads: int = 16
-    hwm: int = 10
-    verbose: bool = False
-
-    @classmethod
-    def from_cache_config(cls, config: CacheConfig) -> FastcacheConfig:
-        cfg_dict = config.model_dump(mode="json", exclude={"pull_uri", "push_uri"})
-        return cls(
-            inurl=config.pull_uri,
-            outurl=config.push_uri,
-            **cfg_dict,
+    def to_fastcache_json(self, indent: int = 2) -> str:
+        """Render the config consumed by the fastcache binary."""
+        data = self.model_dump(
+            mode="json", exclude={"hostname", "pull_uri", "push_uri"}
+        )
+        return json.dumps(
+            {"inurl": str(self.pull_uri), "outurl": str(self.push_uri), **data},
+            indent=indent,
         )
 
 
@@ -73,7 +62,7 @@ class CachePublic(BaseModel):
     id: UUID
     transfer_id: str
     user: str
-    status: CacheStatus
+    state: CacheState
     log_path: Path
     config: CacheConfig
 

src/fastcache_api/process.py

@@ -8,7 +8,7 @@
 import psutil
 
 from .config import settings
-from .models import CacheConfig, CacheProcess, FastcacheConfig
+from .models import CacheConfig, CacheProcess
 
 logger = logging.getLogger(__name__)
 
@@ -40,9 +40,7 @@ def start_cache(cache_id: UUID, config: CacheConfig) -> CacheProcess:
     run_dir.mkdir(parents=True, exist_ok=True)
 
     config_path = run_dir / "config.json"
-    config_path.write_text(
-        FastcacheConfig.from_cache_config(config).model_dump_json(indent=2)
-    )
+    config_path.write_text(config.to_fastcache_json())
 
     log_path = (run_dir / "cache.log").resolve()
     with log_path.open("ab") as log_file:

src/fastcache_api/reconcile.py

@@ -5,26 +5,24 @@
 
 from .config import settings
 from .db import SessionLocal
-from .models import CacheStatus
+from .models import CacheState
 from .process import is_alive
 from .tables import Cache
 
 logger = logging.getLogger(__name__)
 
-# Statuses whose process should still be running, so must be checked vs reality.
-_NON_FINAL = [s.value for s in CacheStatus if not s.is_final()]
+# States whose process should still be running, so must be checked vs reality.
+_NON_FINAL = [s.value for s in CacheState if not s.is_final()]
 
 
 async def sweep_dead_caches() -> int:
     """Mark non-final caches whose process is gone as failed; return how many.
 
-    Caches outlive the api server, so a row's status is a claim we verify against
+    Caches outlive the api server, so a row's state is a claim we verify against
     the live process.
     """
     async with SessionLocal() as session:
-        result = await session.execute(
-            select(Cache).where(Cache.status.in_(_NON_FINAL))
-        )
+        result = await session.execute(select(Cache).where(Cache.state.in_(_NON_FINAL)))
         caches = result.scalars().all()
 
         stale = 0
@@ -36,7 +34,7 @@ async def sweep_dead_caches() -> int:
                 cache.id,
                 cache.pid,
             )
-            cache.status = CacheStatus.failed
+            cache.state = CacheState.failed
             stale += 1
 
         if stale:

src/fastcache_api/routes/cache.py

@@ -14,7 +14,7 @@
     CachePublic,
     CacheRequest,
     CachesPublic,
-    CacheStatus,
+    CacheState,
 )
 from ..process import (
     allocate_port_pair,
@@ -104,7 +104,7 @@ async def create_cache(
         pid=proc.pid,
         create_time=proc.create_time,
         user=req.requested_by,
-        status=CacheStatus.active,
+        state=CacheState.active,
         log_path=str(proc.log_path),
         config=config.model_dump(mode="json"),
     )
@@ -137,7 +137,7 @@ async def shutdown_cache(
             status_code=status.HTTP_404_NOT_FOUND, detail="Cache not found"
         )
 
-    cache.status = CacheStatus.canceled
+    cache.state = CacheState.canceled
     await session.commit()
     await session.refresh(cache)
 

src/fastcache_api/tables.py

@@ -9,7 +9,7 @@
 )
 from sqlalchemy.orm import DeclarativeBase, Mapped, mapped_column
 
-from .models import CacheStatus
+from .models import CacheState
 
 
 class Base(DeclarativeBase):
@@ -46,9 +46,9 @@ class Cache(DTMixin, Base):
         doc=("psutil process create_time at launch. We can key by pid/create_time."),
     )
     user: Mapped[str] = mapped_column(doc="User that created this cache")
-    status: Mapped[str] = mapped_column(
-        default=CacheStatus.new,
-        doc="Lifecycle status; stored as the CacheStatus string value",
+    state: Mapped[str] = mapped_column(
+        default=CacheState.new,
+        doc="Lifecycle state; stored as the CacheState string value",
     )
     log_path: Mapped[str] = mapped_column(
         doc="Absolute path to the cache process log file",