remove SSH to Controller for job status

.env.example

@@ -1,8 +1,4 @@
-CONTROLLER_HOST=ocrd-controller
-CONTROLLER_PORT_SSH=22
-
-MANAGER_DATA=~/.ssh/id_rsa
-MANAGER_KEY=~/
+MANAGER_DATA=~/
 MANAGER_HOST=ocrd-manager
 MANAGER_PORT_WEB=4004
 

Makefile

@@ -28,44 +28,30 @@ Variables:
 	  currently: "$(TAGNAME)"
 	- MONITOR_PORT_WEB	TCP port for the (host-side) web server
 	  currently: $(MONITOR_PORT_WEB)
-	- MANAGER_KEY		SSH key file to mount (for the Controller client)
-	  currently: "$(MANAGER_KEY)"
 	- MANAGER_DATA		host directory to mount into `/data` (shared with Manager)
 	  currently: "$(MANAGER_DATA)"
 	- MANAGER_WORKFLOWS	host directory to mount into `/workflows` (shared with Manager)
 	  currently: "$(MANAGER_WORKFLOWS)"
 	- NETWORK		Docker network to use (manage via "docker network")
 	  currently: $(NETWORK)
-	- CONTROLLER_HOST	network address for the Controller client
-				(must be reachable from the container network)
-	  currently: $(CONTROLLER_HOST)
-	- CONTROLLER_PORT_SSH	network port for the Controller client
-				(must be reachable from the container network)
-	  currently: $(CONTROLLER_PORT_SSH)
 EOF
 endef
 export HELP
 help: ; @eval "$$HELP"
 
-MANAGER_KEY ?= $(firstword $(filter-out %.pub,$(wildcard $(HOME)/.ssh/id_*)))
 MANAGER_DATA ?= $(CURDIR)
 MANAGER_WORKFLOWS ?= $(CURDIR)
 MONITOR_PORT_WEB ?= 5000
 NETWORK ?= bridge
-CONTROLLER_HOST ?= $(shell dig +short $$HOSTNAME)
-CONTROLLER_PORT_SSH ?= 8022
 run: $(DATA)
 	docker run -d --rm \
 	-h ocrd_monitor \
 	--name ocrd_monitor \
 	--network=$(NETWORK) \
 	-p $(MONITOR_PORT_WEB):5000 \
-	-v ${MANAGER_KEY}:/id_rsa \
-	--mount type=bind,source=$(MANAGER_KEY),target=/id_rsa \
 	-v $(MANAGER_DATA):/data \
 	-v $(MANAGER_WORKFLOWS):/workflows \
 	-v shared:/run/lock/ocrd.jobs \
-	-e CONTROLLER=$(CONTROLLER_HOST):$(CONTROLLER_PORT_SSH) \
 	-e MONITOR_PORT_LOG=${MONITOR_PORT_LOG} \
 	$(TAGNAME)
 

README.md

@@ -21,11 +21,8 @@ In order to work properly, the following **environment variables** must be set:
 
 | Variable            | Description                                                                      |
 | ------------------- | -------------------------------------------------------------------------------- |
-| CONTROLLER_HOST     | Hostname of the OCR-D Controller                                                 |
-| CONTROLLER_PORT_SSH | Port on the OCR-D Controller host that allows a SSH connection                   |
 | MANAGER_DATA        | Path to the OCR-D workspaces on the host                                         |
 | MANAGER_WORKFLOWS   | Path to the OCR-D workflows on the host                                          |
-| MANAGER_KEY         | Path to a private key that can be used to authenticate with the OCR-D Controller |
 | MONITOR_PORT_WEB    | The port at which the OCR-D Monitor will be available on the host                |
 | MONITOR_PORT_LOG    | The port at which the Dozzle logs will be available on the host                  |
 

docker-compose.yml

@@ -17,7 +17,6 @@ services:
     hostname: ${MONITOR_HOST}
 
     environment:
-      CONTROLLER: "${CONTROLLER_HOST}:${CONTROLLER_PORT_SSH}"
       MANAGER_URL: "http://${MANAGER_HOST}:${MANAGER_PORT_WEB}"
       MONITOR_PORT_LOG: ${MONITOR_PORT_LOG}
       MONITOR_DB_CONNECTION: "mongodb://${MONITOR_DB_ROOT_USER:-root}:${MONITOR_DB_ROOT_PASSWORD:-root_password}@ocrd-database:27017"
@@ -28,7 +27,6 @@ services:
     volumes:
       - ${MANAGER_DATA}:/data
       - ${MANAGER_WORKFLOWS}:/workflows
-      - ${MANAGER_KEY}:/id_rsa
       - shared:/run/lock/ocrd.jobs
 
   ocrd-logview:

init.sh

@@ -1,31 +1,18 @@
 #!/usr/bin/env bash
 
-mkdir -p ~/.ssh
-cat /id_rsa >> ~/.ssh/id_rsa
-chmod go-rw ~/.ssh/id_rsa
-
-# Add ocrd controller as global and  known_hosts if env exist
-if [ -n "$CONTROLLER" ]; then
-  CONTROLLER_HOST=${CONTROLLER%:*}
-  CONTROLLER_PORT=${CONTROLLER#*:}
-  CONTROLLER_IP=$(nslookup $CONTROLLER_HOST | grep 'Address\:' | awk 'NR==2 {print $2}')
-
-  if test -e /etc/ssh/ssh_known_hosts; then
-    ssh-keygen -R $CONTROLLER_HOST -f /etc/ssh/ssh_known_hosts
-    ssh-keygen -R $CONTROLLER_IP -f /etc/ssh/ssh_known_hosts
-  fi
-  ssh-keyscan -H -p ${CONTROLLER_PORT:-22} $CONTROLLER_HOST,$CONTROLLER_IP >>/etc/ssh/ssh_known_hosts
-fi
-
 export MONITOR_DB_CONNECTION_STRING=$MONITOR_DB_CONNECTION
 export OCRD_BROWSER__MODE=native
-export OCRD_BROWSER__WORKSPACE_DIR=/data/ocr-d
+# all OCR-D workspaces on the Manager are under /data/ocr-d
+# but since the Manager resolves everything under /data
+# it tracks the workspace directory relative to that in the database
+# (e.g. ocr-d/testdata-production)
+# so if we write /data/ocr-d, we could list workspaces fine,
+# but our workspace URLs from the job database would be wrong
+# (resolving as /data/ocr-d/ocr-d/...)
+# so better just use /data as well here:
+export OCRD_BROWSER__WORKSPACE_DIR=/data
 export OCRD_BROWSER__PORT_RANGE="[9000,9100]"
 export OCRD_LOGVIEW__PORT=$MONITOR_PORT_LOG
-export OCRD_CONTROLLER__HOST=$CONTROLLER_HOST
-export OCRD_CONTROLLER__PORT=$CONTROLLER_PORT
-export OCRD_CONTROLLER__USER=admin
-export OCRD_CONTROLLER__KEYFILE=~/.ssh/id_rsa
 export OCRD_MANAGER__URL=$MANAGER_URL
 
 cd /usr/local/ocrd-monitor

ocrdmonitor/database/_browserprocessrepository.py

@@ -29,7 +29,7 @@ def __init__(self, restoring_factory: BrowserRestoringFactory) -> None:
         self._restoring_factory = restoring_factory
 
     async def insert(self, browser: OcrdBrowser) -> None:
-        await BrowserProcess(  # type: ignore
+        await BrowserProcess(
             address=browser.address(),
             owner=browser.owner(),
             process_id=browser.process_id(),

ocrdmonitor/database/_initdb.py

@@ -39,11 +39,11 @@ async def init(connection_str: str, force_initialize: bool = False) -> None:
 
         __initialized = True
         connection_str = rebuild_connection_string(connection_str)
-        client: AsyncIOMotorClient = AsyncIOMotorClient(connection_str) #  type: ignore
-        client.get_io_loop = asyncio.get_event_loop #  type: ignore
+        client = AsyncIOMotorClient(connection_str) # type: ignore[var-annotated]
+        client.get_io_loop = asyncio.get_event_loop # type: ignore[method-assign]
         await init_beanie(
-            database=client.ocrd, #  type: ignore
-            document_models=[BrowserProcess, MongoOcrdJob],  # type: ignore
+            database=client.ocrd,
+            document_models=[BrowserProcess, MongoOcrdJob],
         )
 
     return init

ocrdmonitor/database/_ocrdjobrepository.py

@@ -20,7 +20,6 @@ class MongoOcrdJob(Document):
     workdir: Path
     remotedir: str
     workflow_file: Path
-    controller_address: str
 
     class Settings:
         name = "OcrdJob"
@@ -36,7 +35,7 @@ class Settings:
 
 class MongoJobRepository:
     async def insert(self, job: OcrdJob) -> None:
-        await MongoOcrdJob(**asdict(job)).insert()  # type: ignore
+        await MongoOcrdJob(**asdict(job)).insert()
 
     async def find_all(self) -> list[OcrdJob]:
         return [OcrdJob(**j.dict(exclude={"id"})) for j in await MongoOcrdJob.find_all().to_list()]

ocrdmonitor/environment.py

@@ -9,9 +9,8 @@
     SubProcessOcrdBrowserFactory,
 )
 from ocrdmonitor import database
-from ocrdmonitor.protocols import RemoteServer, Repositories
+from ocrdmonitor.protocols import Repositories
 from ocrdmonitor.server.settings import Settings
-from ocrdmonitor.sshremote import SSHRemote
 
 BrowserType = Type[SubProcessOcrdBrowser] | Type[DockerOcrdBrowser]
 CreatingFactories: dict[str, Callable[[set[int]], OcrdBrowserFactory]] = {
@@ -40,6 +39,3 @@ async def repositories(self) -> Repositories:
     def browser_factory(self) -> OcrdBrowserFactory:
         port_range_set = set(range(*self.settings.ocrd_browser.port_range))
         return CreatingFactories[self.settings.ocrd_browser.mode](port_range_set)
-
-    def controller_server(self) -> RemoteServer:
-        return SSHRemote(self.settings.ocrd_controller)

ocrdmonitor/protocols.py

@@ -4,7 +4,6 @@
 from typing import Collection, NamedTuple, Protocol
 
 from ocrdbrowser import OcrdBrowser, OcrdBrowserFactory
-from ocrdmonitor.processstatus import ProcessStatus
 from ocrdmonitor.server.settings import Settings
 
 
@@ -49,7 +48,6 @@ class OcrdJob:
     workdir: Path
     remotedir: str
     workflow_file: Path
-    controller_address: str
 
     @property
     def is_running(self) -> bool:
@@ -72,13 +70,6 @@ async def find_all(self) -> list[OcrdJob]:
         ...
 
 
-class RemoteServer(Protocol):
-    async def read_file(self, path: str) -> str:
-        ...
-
-    async def process_status(self, process_group: int) -> list[ProcessStatus]:
-        ...
-
 class Repositories(NamedTuple):
     browser_processes: BrowserProcessRepository
     ocrd_jobs: JobRepository
@@ -92,6 +83,3 @@ async def repositories(self) -> Repositories:
 
     def browser_factory(self) -> OcrdBrowserFactory:
         ...
-
-    def controller_server(self) -> RemoteServer:
-        ...

ocrdmonitor/server/jobs.py

@@ -8,20 +8,12 @@
 from fastapi.responses import JSONResponse
 from fastapi.templating import Jinja2Templates
 
-from ocrdmonitor.ocrdcontroller import OcrdController
-from ocrdmonitor.processstatus import ProcessStatus
 from ocrdmonitor.protocols import Environment, OcrdJob, Repositories
 
 import httpx
 import logging
 
 
-@dataclass
-class RunningJob:
-    ocrd_job: OcrdJob
-    process_status: ProcessStatus
-
-
 def split_into_running_and_completed(
     jobs: Iterable[OcrdJob],
 ) -> tuple[list[OcrdJob], list[OcrdJob]]:
@@ -30,25 +22,11 @@ def split_into_running_and_completed(
     return running_ocrd_jobs, completed_ocrd_jobs
 
 
-def wrap_in_running_job_type(
-    running_ocrd_jobs: Iterable[OcrdJob],
-    job_status: Iterable[ProcessStatus | None],
-) -> Iterable[RunningJob]:
-    running_jobs = [
-        RunningJob(job, process_status)
-        for job, process_status in zip(running_ocrd_jobs, job_status)
-        if process_status is not None
-    ]
-
-    return running_jobs
-
-
 def create_jobs(
     templates: Jinja2Templates,
     environment: Environment,
 ) -> APIRouter:
     router = APIRouter(prefix="/jobs")
-    controller = OcrdController(environment.controller_server())
 
     @router.get("/", name="jobs")
     async def jobs(
@@ -58,17 +36,14 @@ async def jobs(
         jobs = await job_repository.find_all()
         running, completed = split_into_running_and_completed(jobs)
 
-        job_status = [await controller.status_for(job) for job in running]
-        running_jobs = wrap_in_running_job_type(running, job_status)
-
         now = datetime.now(timezone.utc)
         return templates.TemplateResponse(
             "jobs.html.j2",
             {
                 "request": request,
                 "running_jobs": sorted(
-                    running_jobs,
-                    key=lambda x: x.ocrd_job.time_created or now,
+                    running,
+                    key=lambda x: x.time_created or now,
                 ),
                 "completed_jobs": sorted(
                     completed,