Readd RBD lock detection and clearing on startup

This is still needed due to the nature of the locks and freeing them on
startup, and to preserve lock=fail behaviour on VM startup.

Also fixes the fencing lock flush to directly use the client library
outside of Celery. I don't like this hack but it seems prudent until we
move fencing to the workers as well.

api-daemon/pvcapid-worker.sh

@@ -25,10 +25,10 @@ CELERY_BIN="$( which celery )"
 # app arguments work in a non-backwards-compatible way with Celery 5.
 case "$( cat /etc/debian_version )" in
     10.*)
-        CELERY_ARGS="worker --app pvcapid.flaskapi.celery --concurrency 1 --hostname $(hostname -s) --queues $(hostname -s) --loglevel INFO"
+        CELERY_ARGS="worker --app pvcapid.flaskapi.celery --concurrency 3 --hostname $(hostname -s) --queues $(hostname -s) --loglevel INFO"
     ;;
     *)
-        CELERY_ARGS="--app pvcapid.flaskapi.celery worker --concurrency 1 --hostname $(hostname -s) --queues $(hostname -s) --loglevel INFO"
+        CELERY_ARGS="--app pvcapid.flaskapi.celery worker --concurrency 3 --hostname $(hostname -s) --queues $(hostname -s) --loglevel INFO"
     ;;
 esac
 

client-cli/pvc/cli/waiters.py

@@ -79,7 +79,7 @@ def wait_for_celery_task(CLI_CONFIG, task_detail):
     # Wait for the task to start
     echo(CLI_CONFIG, "Waiting for task to start...", newline=False)
     while True:
-        sleep(0.25)
+        sleep(0.5)
         task_status = pvc.lib.common.task_status(
             CLI_CONFIG, task_id=task_id, is_watching=True
         )
@@ -104,7 +104,7 @@ def wait_for_celery_task(CLI_CONFIG, task_detail):
             newline=False,
         )
         while True:
-            sleep(0.25)
+            sleep(0.5)
             if task_status.get("state") != "RUNNING":
                 break
             if task_status.get("current") > last_task:

daemon-common/celery.py

@@ -33,10 +33,12 @@ class TaskFailure(Exception):
 def start(celery, msg, current=0, total=1):
     logger = getLogger(__name__)
     logger.info(f"Starting {current}/{total}: {msg}")
+    if celery is None:
+        return
     celery.update_state(
         state="RUNNING", meta={"current": current, "total": total, "status": msg}
     )
-    sleep(0.5)
+    sleep(1)
 
 
 def fail(celery, msg, current=1, total=1):
@@ -64,19 +66,23 @@ def log_err(celery, msg):
 def update(celery, msg, current=1, total=2):
     logger = getLogger(__name__)
     logger.info(f"Task update {current}/{total}: {msg}")
+    if celery is None:
+        return
     celery.update_state(
         state="RUNNING", meta={"current": current, "total": total, "status": msg}
     )
-    sleep(0.5)
+    sleep(1)
 
 
 def finish(celery, msg, current=2, total=2):
     logger = getLogger(__name__)
     logger.info(f"Task update {current}/{total}: Finishing up")
+    if celery is None:
+        return
     celery.update_state(
         state="RUNNING",
         meta={"current": current, "total": total, "status": "Finishing up"},
     )
-    sleep(0.5)
+    sleep(1)
     logger.info(f"Success {current}/{total}: {msg}")
     return {"status": msg, "current": current, "total": total}

node-daemon/plugins/kydb

@@ -0,0 +1,106 @@
+#!/usr/bin/env python3
+
+# kydb.py - PVC Monitoring example plugin for KeyDB/Redis
+# Part of the Parallel Virtual Cluster (PVC) system
+#
+#    Copyright (C) 2018-2023 Joshua M. Boniface <joshua@boniface.me>
+#
+#    This program is free software: you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation, version 3.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#
+#    You should have received a copy of the GNU General Public License
+#    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+#
+###############################################################################
+
+# This script provides an example of a PVC monitoring plugin script. It will create
+# a simple plugin to check the Libvirt daemon instance on the node for operation.
+
+# This script can thus be used as an example or reference implementation of a
+# PVC monitoring pluginscript and expanded upon as required.
+
+# A monitoring plugin script must implement the class "MonitoringPluginScript" which
+# extends "MonitoringPlugin", providing the 3 functions indicated. Detailed explanation
+# of the role of each function is provided in context of the example; see the other
+# examples for more potential uses.
+
+# WARNING:
+#
+# This script will run in the context of the node daemon keepalives as root.
+# DO NOT install untrusted, unvetted plugins under any circumstances.
+
+
+# This import is always required here, as MonitoringPlugin is used by the
+# MonitoringPluginScript class
+from pvcnoded.objects.MonitoringInstance import MonitoringPlugin
+
+
+# A monitoring plugin script must always expose its nice name, which must be identical to
+# the file name
+PLUGIN_NAME = "kydb"
+
+
+# The MonitoringPluginScript class must be named as such, and extend MonitoringPlugin.
+class MonitoringPluginScript(MonitoringPlugin):
+    def setup(self):
+        """
+        setup(): Perform special setup steps during node daemon startup
+
+        This step is optional and should be used sparingly.
+
+        If you wish for the plugin to not kydb in certain conditions, do any checks here
+        and return a non-None failure message to indicate the error.
+        """
+
+        pass
+
+    def run(self, coordinator_state=None):
+        """
+        run(): Perform the check actions and return a PluginResult object
+
+        The {coordinator_state} can be used to check if this is a "primary" coordinator, "secondary" coordinator, or "client" (non-coordinator)
+        """
+
+        # Run any imports first
+        from redis import Redis
+
+        rd_conn = None
+
+        # Set the health delta to 0 (no change)
+        health_delta = 0
+        # Craft a message that can be used by the clients
+        message = "Successfully connected to Libvirtd on localhost"
+
+        # Check the Zookeeper connection
+        try:
+            rd_conn = Redis(host='localhost', port=6379, decode_responses=True)
+            data = rd_conn.info()
+        except Exception as e:
+            health_delta = 50
+            message = f"Failed to connect to KeyDB/Redis: {e}"
+        finally:
+            del rd_conn
+
+        # Set the health delta in our local PluginResult object
+        self.plugin_result.set_health_delta(health_delta)
+
+        # Set the message in our local PluginResult object
+        self.plugin_result.set_message(message)
+
+        # Return our local PluginResult object
+        return self.plugin_result
+
+    def cleanup(self):
+        """
+        cleanup(): Perform special cleanup steps during node daemon termination
+
+        This step is optional and should be used sparingly.
+        """
+
+        pass

node-daemon/pvcnoded/objects/VMInstance.py

@@ -24,8 +24,8 @@ import time
 import libvirt
 
 from threading import Thread
-
 from xml.etree import ElementTree
+from json import loads as jloads
 
 import daemon_lib.common as common
 
@@ -283,9 +283,49 @@ class VMInstance(object):
             self.logger.out(
                 "Flushing RBD locks", state="i", prefix="Domain {}".format(self.domuuid)
             )
-            VMInstance.flush_locks(
+
-                self.zkhandler, self.logger, self.domuuid, self.this_node
+            rbd_list = self.zkhandler.read(
+                ("domain.storage.volumes", self.domuuid)
+            ).split(",")
+
+            locks = list()
+            for rbd in rbd_list:
+                retcode, stdout, stderr = common.run_os_command(
+                    f"rbd lock list --format json {rbd}"
                 )
+                if retcode == 0:
+                    _locks = jloads(stdout)
+                    for lock in _locks:
+                        lock["rbd"] = rbd
+                        locks.append(lock)
+
+            for lock in locks:
+                lockid = lock["id"]
+                locker = lock["locker"]
+                owner = lock["address"].split(":")[0]
+                rbd = lock["rbd"]
+
+                if owner == self.this_node.storage_ipaddr:
+                    retcode, stdout, stderr = common.run_os_command(
+                        f'rbd lock remove {rbd} "{lockid}" "{locker}"'
+                    )
+                else:
+                    self.logger.out(
+                        "RBD lock does not belong to this host (owner {owner}) so freeing this long is dangerous; aborting VM start",
+                        state="e",
+                        prefix="Domain {}".format(self.domuuid),
+                    )
+                    self.zkhandler.write(
+                        [
+                            (("domain.state", self.domuuid), "fail"),
+                            (
+                                ("domain.failed_reason", self.domuuid),
+                                f"Could not safely free RBD lock {lockid} ({owner}) on volume {rbd}; stop VM and flush locks manually",
+                            ),
+                        ]
+                    )
+                    break
+
             if self.zkhandler.read(("domain.state", self.domuuid)) == "fail":
                 lv_conn.close()
                 self.dom = None

node-daemon/pvcnoded/util/fencing.py

@@ -23,7 +23,7 @@ import time
 
 import daemon_lib.common as common
 
-from pvcnoded.objects.VMInstance import VMInstance
+from daemon_lib.vm import vm_worker_flush_locks
 
 
 #
@@ -121,7 +121,7 @@ def migrateFromFencedNode(zkhandler, node_name, config, logger):
 
     # Migrate a VM after a flush
     def fence_migrate_vm(dom_uuid):
-        VMInstance.flush_locks(zkhandler, logger, dom_uuid)
+        vm_worker_flush_locks(zkhandler, None, dom_uuid, force_unlock=True)
 
         target_node = common.findTargetNode(zkhandler, dom_uuid)