Add is-primary command

Used by the cron to check if the node firing an autobackup is the
primary node or not, so it will not multi-fire from all coordinators.

api-daemon/pvcapid/flaskapi.py

@@ -3515,6 +3515,74 @@ class API_VM_Snapshot_Import(Resource):
 api.add_resource(API_VM_Snapshot_Import, "/vm/<vm>/snapshot/import")
 
 
+# /vm/autobackup
+class API_VM_Autobackup_Root(Resource):
+    @RequestParser(
+        [
+            {"name": "force_full"},
+            {"name": "email_recipients"},
+        ]
+    )
+    @Authenticator
+    def post(self, reqargs):
+        """
+        Trigger a cluster autobackup job
+        ---
+        tags:
+          - provisioner
+        parameters:
+          - in: query
+            name: force_full
+            type: boolean
+            required: false
+            description: If set and true, triggers a full autobackup regardless of schedule
+          - in: query
+            name: email_recipients
+            type: array
+            description: A list of email addresses to send failure and report emails to
+            items:
+              type: string
+              example: "user@domain.tld"
+        responses:
+          200:
+            description: OK
+            schema:
+              type: object
+              properties:
+                task_id:
+                  type: string
+                  description: Task ID for the provisioner Celery worker
+          400:
+            description: Bad request
+            schema:
+              type: object
+              id: Message
+        """
+
+        email_recipients = reqargs.get("email_recipients", None)
+        if email_recipients is not None and not isinstance(email_recipients, list):
+            email_recipients = [email_recipients]
+
+        task = run_celery_task(
+            "cluster.autobackup",
+            force_full=bool(strtobool(reqargs.get("force_full", "false"))),
+            email_recipients=email_recipients,
+            run_on="primary",
+        )
+        return (
+            {
+                "task_id": task.id,
+                "task_name": "cluster.autobackup",
+                "run_on": f"{get_primary_node()} (primary)",
+            },
+            202,
+            {"Location": Api.url_for(api, API_Tasks_Element, task_id=task.id)},
+        )
+
+
+api.add_resource(API_VM_Autobackup_Root, "/vm/autobackup")
+
+
 ##########################################################
 # Client API - Network
 ##########################################################
@@ -5070,7 +5138,7 @@ class API_Storage_Ceph_Benchmark(Resource):
             {
                 "task_id": task.id,
                 "task_name": "storage.benchmark",
-                "run_on": get_primary_node(),
+                "run_on": f"{get_primary_node()} (primary)",
             },
             202,
             {"Location": Api.url_for(api, API_Tasks_Element, task_id=task.id)},
@@ -9326,7 +9394,7 @@ class API_Provisioner_Create_Root(Resource):
             {
                 "task_id": task.id,
                 "task_name": "provisioner.create",
-                "run_on": get_primary_node(),
+                "run_on": f"{get_primary_node()} (primary)",
             },
             202,
             {"Location": Api.url_for(api, API_Tasks_Element, task_id=task.id)},

build-and-deploy.sh

@@ -13,6 +13,7 @@ else
 fi
 
 KEEP_ARTIFACTS=""
+API_ONLY=""
 PRIMARY_NODE=""
 if [[ -n ${1} ]]; then
     for arg in ${@}; do
@@ -21,6 +22,10 @@ if [[ -n ${1} ]]; then
                 KEEP_ARTIFACTS="y"
                 shift
             ;;
+            -a|--api-only)
+                API_ONLY="y"
+                shift
+            ;;
             -p=*|--become-primary=*)
                 PRIMARY_NODE=$( awk -F'=' '{ print $NF }' <<<"${arg}" )
                 shift
@@ -75,6 +80,7 @@ for HOST in ${HOSTS[@]}; do
     ssh $HOST $SUDO systemctl restart pvcapid &>/dev/null
     sleep 2
     ssh $HOST $SUDO systemctl restart pvcworkerd &>/dev/null
+    if [[ -z ${API_ONLY} ]]; then
     sleep 2
     ssh $HOST $SUDO systemctl restart pvchealthd &>/dev/null
     sleep 2
@@ -85,10 +91,13 @@ for HOST in ${HOSTS[@]}; do
         sleep 5
         echo -n "."
     done
+    fi
     echo " done."
     if [[ -n ${PRIMARY_NODE} && ${PRIMARY_NODE} == ${HOST} ]]; then
-        echo -n ">>> "
-        ssh $HOST pvc -q node primary
+        echo -n ">>> Setting node $HOST to primary coordinator state... "
+        ssh $HOST pvc -q node primary --wait &>/dev/null
+        ssh $HOST $SUDO systemctl restart pvcworkerd &>/dev/null
+        echo "done."
     fi
 done
 

client-cli/pvc/cli/cli.py

@@ -724,6 +724,33 @@ def cli_node():
     pass
 
 
+###############################################################################
+# > pvc node is-primary
+###############################################################################
+@click.command(
+    name="is-primary",
+    short_help="Check if this node is primary coordinator.",
+)
+@connection_req
+@click.argument("node", default=DEFAULT_NODE_HOSTNAME)
+def cli_node_is_primary(
+    node,
+):
+    """
+    Check if NODE (or this node if unset) is the current primary coordinator.
+
+    Designed for scripting; returns no visible data, but the return code is 0 if the node
+    is primary, and 1 if it is not.
+    """
+
+    _, primary_node = pvc.lib.cluster.get_primary_node(CLI_CONFIG)
+
+    if primary_node == node:
+        exit(0)
+    else:
+        exit(1)
+
+
 ###############################################################################
 # > pvc node primary
 ###############################################################################
@@ -1749,7 +1776,7 @@ def cli_vm_unmigrate(domain, wait, force_live):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 def cli_vm_flush_locks(domain, wait_flag):
     """
@@ -1793,7 +1820,7 @@ def cli_vm_snapshot():
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 def cli_vm_snapshot_create(domain, snapshot_name, wait_flag):
     """
@@ -1827,7 +1854,7 @@ def cli_vm_snapshot_create(domain, snapshot_name, wait_flag):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt("Remove shapshot {snapshot_name} of VM {domain}")
 def cli_vm_snapshot_remove(domain, snapshot_name, wait_flag):
@@ -1860,7 +1887,7 @@ def cli_vm_snapshot_remove(domain, snapshot_name, wait_flag):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt(
     "Roll back to snapshot {snapshot_name} of {domain} and lose all data and changes since this snapshot"
@@ -1903,7 +1930,7 @@ def cli_vm_snapshot_rollback(domain, snapshot_name, wait_flag):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 def cli_vm_snapshot_export(
     domain, snapshot_name, export_path, incremental_parent, wait_flag
@@ -1957,7 +1984,7 @@ def cli_vm_snapshot_export(
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 def cli_vm_snapshot_import(
     domain, snapshot_name, import_path, retain_snapshot, wait_flag
@@ -2001,6 +2028,8 @@ def cli_vm_snapshot_import(
 )
 def cli_vm_backup():
     """
+    DEPRECATED: Use 'pvc vm snapshot' commands instead. 'pvc vm backup' commands will be removed in PVC 0.9.100.
+
     Manage backups of VMs in a PVC cluster.
     """
     pass
@@ -2030,6 +2059,8 @@ def cli_vm_backup():
 )
 def cli_vm_backup_create(domain, backup_path, incremental_parent, retain_snapshot):
     """
+    DEPRECATED: Use 'pvc vm snapshot' commands instead. 'pvc vm backup' commands will be removed in PVC 0.9.100.
+
     Create a backup of virtual machine DOMAIN to BACKUP_PATH on the cluster primary coordinator. DOMAIN may be a UUID or name.
 
     BACKUP_PATH must be a valid absolute directory path on the cluster "primary" coordinator (see "pvc node list") allowing writes from the API daemon (normally running as "root"). The BACKUP_PATH should be a large storage volume, ideally a remotely mounted filesystem (e.g. NFS, SSHFS, etc.) or non-Ceph-backed disk; PVC does not handle this path, that is up to the administrator to configure and manage.
@@ -2076,6 +2107,8 @@ def cli_vm_backup_create(domain, backup_path, incremental_parent, retain_snapsho
 )
 def cli_vm_backup_restore(domain, backup_datestring, backup_path, retain_snapshot):
     """
+    DEPRECATED: Use 'pvc vm snapshot' commands instead. 'pvc vm backup' commands will be removed in PVC 0.9.100.
+
     Restore the backup BACKUP_DATESTRING of virtual machine DOMAIN stored in BACKUP_PATH on the cluster primary coordinator. DOMAIN may be a UUID or name.
 
     BACKUP_PATH must be a valid absolute directory path on the cluster "primary" coordinator (see "pvc node list") allowing reads from the API daemon (normally running as "root"). The BACKUP_PATH should be a large storage volume, ideally a remotely mounted filesystem (e.g. NFS, SSHFS, etc.) or non-Ceph-backed disk; PVC does not handle this path, that is up to the administrator to configure and manage.
@@ -2114,6 +2147,8 @@ def cli_vm_backup_restore(domain, backup_datestring, backup_path, retain_snapsho
 @click.argument("backup_path")
 def cli_vm_backup_remove(domain, backup_datestring, backup_path):
     """
+    DEPRECATED: Use 'pvc vm snapshot' commands instead. 'pvc vm backup' commands will be removed in PVC 0.9.100.
+
     Remove the backup BACKUP_DATESTRING, including snapshots, of virtual machine DOMAIN stored in BACKUP_PATH on the cluster primary coordinator. DOMAIN may be a UUID or name.
 
     WARNING: Removing an incremental parent will invalidate any existing incremental backups based on that backup.
@@ -2141,15 +2176,6 @@ def cli_vm_backup_remove(domain, backup_datestring, backup_path):
     name="autobackup", short_help="Perform automatic virtual machine backups."
 )
 @connection_req
-@click.option(
-    "-f",
-    "--configuration",
-    "autobackup_cfgfile",
-    envvar="PVC_AUTOBACKUP_CFGFILE",
-    default=DEFAULT_AUTOBACKUP_FILENAME,
-    show_default=True,
-    help="Override default config file location.",
-)
 @click.option(
     "--email-report",
     "email_report",
@@ -2163,39 +2189,42 @@ def cli_vm_backup_remove(domain, backup_datestring, backup_path):
     is_flag=True,
     help="Force all backups to be full backups this run.",
 )
+@click.option(
+    "--wait/--no-wait",
+    "wait_flag",
+    is_flag=True,
+    default=True,
+    show_default=True,
+    help="Wait or don't wait for task to complete, showing progress if waiting.",
+)
 @click.option(
     "--cron",
     "cron_flag",
-    default=False,
     is_flag=True,
-    help="Cron mode; don't error exit if this isn't the primary coordinator.",
+    default=False,
+    show_default=True,
+    help="Run in cron mode (returns immediately with no output once job is submitted).",
 )
-def cli_vm_autobackup(autobackup_cfgfile, email_report, force_full_flag, cron_flag):
+def cli_vm_autobackup(email_report, force_full_flag, wait_flag, cron_flag):
     """
     Perform automated backups of VMs, with integrated cleanup and full/incremental scheduling.
 
-    This command enables automatic backup of PVC VMs at the block level, leveraging the various "pvc vm backup"
+    This command enables automatic backup of PVC VMs at the block level, leveraging the various "pvc vm snapshot"
     functions with an internal rentention and cleanup system as well as determination of full vs. incremental
     backups at different intervals. VMs are selected based on configured VM tags. The destination storage
     may either be local, or provided by a remote filesystem which is automatically mounted and unmounted during
     the backup run via a set of configured commands before and after the backup run.
 
-    NOTE: This command performs its tasks in a local context. It MUST be run from the cluster's active primary
-    coordinator using the "local" connection only; if either is not correct, the command will error.
-
-    NOTE: This command should be run as the same user as the API daemon, usually "root" with "sudo -E" or in
-    a cronjob as "root", to ensure permissions are correct on the backup files. Failure to do so will still take
-    the backup, but the state update write will likely fail and the backup will become untracked. The command
-    will prompt for confirmation if it is found not to be running as "root" and this cannot be bypassed.
-
     This command should be run from cron or a timer at a regular interval (e.g. daily, hourly, etc.) which defines
     how often backups are taken. Backup format (full/incremental) and retention is based only on the number of
-    recorded backups, not on the time interval between them. Backups taken manually outside of the "autobackup"
+    recorded backups, not on the time interval between them. Exports taken manually outside of the "autobackup"
     command are not counted towards the format or retention of autobackups.
 
-    The PVC_AUTOBACKUP_CFGFILE envvar or "-f"/"--configuration" option can be used to override the default
-    configuration file path if required by a particular run. For full details of the possible options, please
-    see the example configuration file at "/usr/share/pvc/autobackup.sample.yaml".
+    WARNING: Running this command manually will interfere with the schedule! Do not run manually except for testing.
+
+    The actual details of the autobackup, including retention policies, full-vs-incremental, pre- and post- run
+    mounting/unmounting commands, etc. are defined in the main PVC configuration file `/etc/pvc/pvc.conf`. See
+    the sample configuration for more details.
 
     An optional report on all current backups can be emailed to one or more email addresses using the
     "--email-report" flag. This report will include information on all current known backups.
@@ -2204,11 +2233,29 @@ def cli_vm_autobackup(autobackup_cfgfile, email_report, force_full_flag, cron_fl
     which can help synchronize the backups of existing VMs with new ones.
     """
 
-    # All work here is done in the helper function for portability; we don't even use "finish"
-    vm_autobackup(
-        CLI_CONFIG, autobackup_cfgfile, email_report, force_full_flag, cron_flag
+    if cron_flag:
+        wait_flag = False
+
+    if email_report is not None:
+        email_recipients = email_report.split(",")
+    else:
+        email_recipients = None
+
+    retcode, retmsg = pvc.lib.vm.vm_autobackup(
+        CLI_CONFIG,
+        email_recipients=email_recipients,
+        force_full_flag=force_full_flag,
+        wait_flag=wait_flag,
     )
 
+    if retcode and wait_flag:
+        retmsg = wait_for_celery_task(CLI_CONFIG, retmsg)
+
+    if cron_flag:
+        finish(retcode, None)
+    else:
+        finish(retcode, retmsg)
+
 
 ###############################################################################
 # > pvc vm tag
@@ -3714,7 +3761,7 @@ def cli_storage_benchmark():
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt(
     "Storage benchmarks take approximately 10 minutes to run and generate significant load on the cluster; they should be run sparingly. Continue"
@@ -3803,7 +3850,7 @@ def cli_storage_osd():
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt(
     "Destroy all data on and create a new OSD database volume group on node {node} device {device}"
@@ -3878,7 +3925,7 @@ def cli_storage_osd_create_db_vg(node, device, wait_flag):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt("Destroy all data on and create new OSD(s) on node {node} device {device}")
 def cli_storage_osd_add(
@@ -3961,7 +4008,7 @@ def cli_storage_osd_add(
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt(
     "Destroy all data on and replace OSD {osdid} (and peer split OSDs) with new device {new_device}"
@@ -4016,7 +4063,7 @@ def cli_storage_osd_replace(
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt("Refresh OSD {osdid} (and peer split OSDs) on device {device}")
 def cli_storage_osd_refresh(osdid, device, wait_flag):
@@ -4061,7 +4108,7 @@ def cli_storage_osd_refresh(osdid, device, wait_flag):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 @confirm_opt("Remove and destroy data on OSD {osdid}")
 def cli_storage_osd_remove(osdid, force_flag, wait_flag):
@@ -6113,7 +6160,7 @@ def cli_provisioner_profile_list(limit, format_function):
     is_flag=True,
     default=True,
     show_default=True,
-    help="Wait or don't wait for task to complete, showing progress",
+    help="Wait or don't wait for task to complete, showing progress if waiting",
 )
 def cli_provisioner_create(
     name, profile, define_flag, start_flag, script_args, wait_flag
@@ -6505,6 +6552,7 @@ def cli(
 # Click command tree
 ###############################################################################
 
+cli_node.add_command(cli_node_is_primary)
 cli_node.add_command(cli_node_primary)
 cli_node.add_command(cli_node_secondary)
 cli_node.add_command(cli_node_flush)

client-cli/pvc/cli/helpers.py

@@ -20,26 +20,16 @@
 ###############################################################################
 
 from click import echo as click_echo
-from click import confirm
-from datetime import datetime
 from distutils.util import strtobool
-from getpass import getuser
 from json import load as jload
 from json import dump as jdump
-from os import chmod, environ, getpid, path, popen, makedirs, get_terminal_size
-from re import findall
+from os import chmod, environ, getpid, path, get_terminal_size
 from socket import gethostname
-from subprocess import run, PIPE
 from sys import argv
 from syslog import syslog, openlog, closelog, LOG_AUTH
 from yaml import load as yload
 from yaml import SafeLoader
 
-import pvc.lib.provisioner
-import pvc.lib.vm
-import pvc.lib.node
-import pvc.lib.storage
-
 
 DEFAULT_STORE_DATA = {"cfgfile": "/etc/pvc/pvc.conf"}
 DEFAULT_STORE_FILENAME = "pvc.json"
@@ -196,452 +186,3 @@ def update_store(store_path, store_data):
 
     with open(store_file, "w") as fh:
         jdump(store_data, fh, sort_keys=True, indent=4)
-
-
-def get_autobackup_config(CLI_CONFIG, cfgfile):
-    try:
-        config = dict()
-        with open(cfgfile) as fh:
-            full_config = yload(fh, Loader=SafeLoader)
-        backup_config = full_config["autobackup"]
-        config["backup_root_path"] = backup_config["backup_root_path"]
-        config["backup_root_suffix"] = backup_config["backup_root_suffix"]
-        config["backup_tags"] = backup_config["backup_tags"]
-        config["backup_schedule"] = backup_config["backup_schedule"]
-        config["auto_mount_enabled"] = backup_config["auto_mount"]["enabled"]
-        if config["auto_mount_enabled"]:
-            config["mount_cmds"] = list()
-            _mount_cmds = backup_config["auto_mount"]["mount_cmds"]
-            for _mount_cmd in _mount_cmds:
-                if "{backup_root_path}" in _mount_cmd:
-                    _mount_cmd = _mount_cmd.format(
-                        backup_root_path=backup_config["backup_root_path"]
-                    )
-                config["mount_cmds"].append(_mount_cmd)
-
-            config["unmount_cmds"] = list()
-            _unmount_cmds = backup_config["auto_mount"]["unmount_cmds"]
-            for _unmount_cmd in _unmount_cmds:
-                if "{backup_root_path}" in _unmount_cmd:
-                    _unmount_cmd = _unmount_cmd.format(
-                        backup_root_path=backup_config["backup_root_path"]
-                    )
-                config["unmount_cmds"].append(_unmount_cmd)
-    except FileNotFoundError:
-        return "Backup configuration does not exist!"
-    except KeyError as e:
-        return f"Backup configuration is invalid: {e}"
-
-    return config
-
-
-def vm_autobackup(
-    CLI_CONFIG,
-    autobackup_cfgfile=DEFAULT_AUTOBACKUP_FILENAME,
-    email_report=None,
-    force_full_flag=False,
-    cron_flag=False,
-):
-    """
-    Perform automatic backups of VMs based on an external config file.
-    """
-
-    backup_summary = dict()
-
-    if email_report is not None:
-        from email.utils import formatdate
-        from socket import gethostname
-
-        try:
-            with open(autobackup_cfgfile) as fh:
-                tmp_config = yload(fh, Loader=SafeLoader)
-            cluster = tmp_config["cluster"]["name"]
-        except Exception:
-            cluster = "unknown"
-
-    def send_execution_failure_report(error=None):
-        echo(CLI_CONFIG, f"Sending email failure report to {email_report}")
-
-        current_datetime = datetime.now()
-        email_datetime = formatdate(float(current_datetime.strftime("%s")))
-
-        email = list()
-        email.append(f"Date: {email_datetime}")
-        email.append(f"Subject: PVC Autobackup execution failure for cluster {cluster}")
-
-        recipients = list()
-        for recipient in email_report.split(","):
-            recipients.append(f"<{recipient}>")
-        email.append(f"To: {', '.join(recipients)}")
-        email.append(f"From: PVC Autobackup System <pvc@{gethostname()}>")
-        email.append("")
-
-        email.append(
-            f"A PVC autobackup has FAILED at {current_datetime} due to an execution error."
-        )
-        email.append("")
-        email.append("The reported error message is:")
-        email.append(f"    {error}")
-
-        try:
-            p = popen("/usr/sbin/sendmail -t", "w")
-            p.write("\n".join(email))
-            p.close()
-        except Exception as e:
-            echo(CLI_CONFIG, f"Failed to send report email: {e}")
-
-    # Validate that we are running on the current primary coordinator of the 'local' cluster connection
-    real_connection = CLI_CONFIG["connection"]
-    CLI_CONFIG["connection"] = "local"
-    retcode, retdata = pvc.lib.node.node_info(CLI_CONFIG, DEFAULT_NODE_HOSTNAME)
-    if not retcode or retdata.get("coordinator_state") != "primary":
-        if cron_flag:
-            echo(
-                CLI_CONFIG,
-                "Current host is not the primary coordinator of the local cluster and running in cron mode. Exiting cleanly.",
-            )
-            exit(0)
-        else:
-            echo(
-                CLI_CONFIG,
-                f"ERROR: Current host is not the primary coordinator of the local cluster; got connection '{real_connection}', host '{DEFAULT_NODE_HOSTNAME}'.",
-            )
-            echo(
-                CLI_CONFIG,
-                "Autobackup MUST be run from the cluster active primary coordinator using the 'local' connection. See '-h'/'--help' for details.",
-            )
-            if email_report is not None:
-                send_execution_failure_report(
-                    error=f"Autobackup run attempted from non-local connection or non-primary coordinator; got connection '{real_connection}', host '{DEFAULT_NODE_HOSTNAME}'."
-                )
-            exit(1)
-
-    # Ensure we're running as root, or show a warning & confirmation
-    if getuser() != "root":
-        confirm(
-            "WARNING: You are not running this command as 'root'. This command should be run under the same user as the API daemon, which is usually 'root'. Are you sure you want to continue?",
-            prompt_suffix=" ",
-            abort=True,
-        )
-
-    # Load our YAML config
-    autobackup_config = get_autobackup_config(CLI_CONFIG, autobackup_cfgfile)
-    if not isinstance(autobackup_config, dict):
-        echo(CLI_CONFIG, f"ERROR: {autobackup_config}")
-        if email_report is not None:
-            send_execution_failure_report(error=f"{autobackup_config}")
-        exit(1)
-
-    # Get the start time of this run
-    autobackup_start_time = datetime.now()
-
-    # Get a list of all VMs on the cluster
-    # We don't do tag filtering here, because we could match an arbitrary number of tags; instead, we
-    # parse the list after
-    retcode, retdata = pvc.lib.vm.vm_list(CLI_CONFIG, None, None, None, None, None)
-    if not retcode:
-        echo(CLI_CONFIG, f"ERROR: Failed to fetch VM list: {retdata}")
-        if email_report is not None:
-            send_execution_failure_report(error=f"Failed to fetch VM list: {retdata}")
-        exit(1)
-    cluster_vms = retdata
-
-    # Parse the list to match tags; too complex for list comprehension alas
-    backup_vms = list()
-    for vm in cluster_vms:
-        vm_tag_names = [t["name"] for t in vm["tags"]]
-        matching_tags = (
-            True
-            if len(
-                set(vm_tag_names).intersection(set(autobackup_config["backup_tags"]))
-            )
-            > 0
-            else False
-        )
-        if matching_tags:
-            backup_vms.append(vm["name"])
-
-    if len(backup_vms) < 1:
-        echo(CLI_CONFIG, "Found no suitable VMs for autobackup.")
-        exit(0)
-
-    # Pretty print the names of the VMs we'll back up (to stderr)
-    maxnamelen = max([len(n) for n in backup_vms]) + 2
-    cols = 1
-    while (cols * maxnamelen + maxnamelen + 2) <= MAX_CONTENT_WIDTH:
-        cols += 1
-    rows = len(backup_vms) // cols
-    vm_list_rows = list()
-    for row in range(0, rows + 1):
-        row_start = row * cols
-        row_end = (row * cols) + cols
-        row_str = ""
-        for x in range(row_start, row_end):
-            if x < len(backup_vms):
-                row_str += "{:<{}}".format(backup_vms[x], maxnamelen)
-        vm_list_rows.append(row_str)
-
-    echo(CLI_CONFIG, f"Found {len(backup_vms)} suitable VM(s) for autobackup.")
-    echo(CLI_CONFIG, "Full VM list:", stderr=True)
-    echo(CLI_CONFIG, "  {}".format("\n  ".join(vm_list_rows)), stderr=True)
-    echo(CLI_CONFIG, "", stderr=True)
-
-    if autobackup_config["auto_mount_enabled"]:
-        # Execute each mount_cmds command in sequence
-        for cmd in autobackup_config["mount_cmds"]:
-            echo(
-                CLI_CONFIG,
-                f"Executing mount command '{cmd.split()[0]}'... ",
-                newline=False,
-            )
-            tstart = datetime.now()
-            ret = run(
-                cmd.split(),
-                stdout=PIPE,
-                stderr=PIPE,
-            )
-            tend = datetime.now()
-            ttot = tend - tstart
-            if ret.returncode != 0:
-                echo(
-                    CLI_CONFIG,
-                    f"failed. [{ttot.seconds}s]",
-                )
-                echo(
-                    CLI_CONFIG,
-                    f"Exiting; command reports: {ret.stderr.decode().strip()}",
-                )
-                if email_report is not None:
-                    send_execution_failure_report(error=ret.stderr.decode().strip())
-                exit(1)
-            else:
-                echo(CLI_CONFIG, f"done. [{ttot.seconds}s]")
-
-    # For each VM, perform the backup
-    for vm in backup_vms:
-        backup_suffixed_path = f"{autobackup_config['backup_root_path']}{autobackup_config['backup_root_suffix']}"
-        if not path.exists(backup_suffixed_path):
-            makedirs(backup_suffixed_path)
-
-        backup_path = f"{backup_suffixed_path}/{vm}"
-        autobackup_state_file = f"{backup_path}/.autobackup.json"
-        if not path.exists(backup_path) or not path.exists(autobackup_state_file):
-            # There are no new backups so the list is empty
-            state_data = dict()
-            tracked_backups = list()
-        else:
-            with open(autobackup_state_file) as fh:
-                state_data = jload(fh)
-            tracked_backups = state_data["tracked_backups"]
-
-        full_interval = autobackup_config["backup_schedule"]["full_interval"]
-        full_retention = autobackup_config["backup_schedule"]["full_retention"]
-
-        full_backups = [b for b in tracked_backups if b["type"] == "full"]
-        if len(full_backups) > 0:
-            last_full_backup = full_backups[0]
-            last_full_backup_idx = tracked_backups.index(last_full_backup)
-            if force_full_flag:
-                this_backup_type = "forced-full"
-                this_backup_incremental_parent = None
-                this_backup_retain_snapshot = True
-            elif last_full_backup_idx >= full_interval - 1:
-                this_backup_type = "full"
-                this_backup_incremental_parent = None
-                this_backup_retain_snapshot = True
-            else:
-                this_backup_type = "incremental"
-                this_backup_incremental_parent = last_full_backup["datestring"]
-                this_backup_retain_snapshot = False
-        else:
-            # The very first backup must be full to start the tree
-            this_backup_type = "full"
-            this_backup_incremental_parent = None
-            this_backup_retain_snapshot = True
-
-        # Perform the backup
-        echo(
-            CLI_CONFIG,
-            f"Backing up VM '{vm}' ({this_backup_type})... ",
-            newline=False,
-        )
-        tstart = datetime.now()
-        retcode, retdata = pvc.lib.vm.vm_backup(
-            CLI_CONFIG,
-            vm,
-            backup_suffixed_path,
-            incremental_parent=this_backup_incremental_parent,
-            retain_snapshot=this_backup_retain_snapshot,
-        )
-        tend = datetime.now()
-        ttot = tend - tstart
-        if not retcode:
-            backup_datestring = findall(r"[0-9]{14}", retdata)[0]
-            echo(CLI_CONFIG, f"failed. [{ttot.seconds}s]")
-            echo(
-                CLI_CONFIG,
-                retdata.strip().replace(f"ERROR in backup {backup_datestring}: ", ""),
-            )
-            skip_cleanup = True
-        else:
-            backup_datestring = findall(r"[0-9]{14}", retdata)[0]
-            echo(
-                CLI_CONFIG,
-                f"done. Backup '{backup_datestring}' created. [{ttot.seconds}s]",
-            )
-            skip_cleanup = False
-
-        # Read backup file to get details
-        backup_json_file = f"{backup_path}/{backup_datestring}/pvcbackup.json"
-        with open(backup_json_file) as fh:
-            backup_json = jload(fh)
-        tracked_backups.insert(0, backup_json)
-
-        # Delete any full backups that are expired
-        marked_for_deletion = list()
-        found_full_count = 0
-        for backup in tracked_backups:
-            if backup["type"] == "full":
-                found_full_count += 1
-                if found_full_count > full_retention:
-                    marked_for_deletion.append(backup)
-
-        # Depete any incremental backups that depend on marked parents
-        for backup in tracked_backups:
-            if backup["type"] == "incremental" and backup["incremental_parent"] in [
-                b["datestring"] for b in marked_for_deletion
-            ]:
-                marked_for_deletion.append(backup)
-
-        if len(marked_for_deletion) > 0:
-            if skip_cleanup:
-                echo(
-                    CLI_CONFIG,
-                    f"Skipping cleanups for {len(marked_for_deletion)} aged-out backups due to backup failure.",
-                )
-            else:
-                echo(
-                    CLI_CONFIG,
-                    f"Running cleanups for {len(marked_for_deletion)} aged-out backups...",
-                )
-                # Execute deletes
-                for backup_to_delete in marked_for_deletion:
-                    echo(
-                        CLI_CONFIG,
-                        f"Removing old VM '{vm}' backup '{backup_to_delete['datestring']}' ({backup_to_delete['type']})... ",
-                        newline=False,
-                    )
-                    tstart = datetime.now()
-                    retcode, retdata = pvc.lib.vm.vm_remove_backup(
-                        CLI_CONFIG,
-                        vm,
-                        backup_suffixed_path,
-                        backup_to_delete["datestring"],
-                    )
-                    tend = datetime.now()
-                    ttot = tend - tstart
-                    if not retcode:
-                        echo(CLI_CONFIG, f"failed. [{ttot.seconds}s]")
-                        echo(
-                            CLI_CONFIG,
-                            f"Skipping removal from tracked backups; command reports: {retdata}",
-                        )
-                    else:
-                        tracked_backups.remove(backup_to_delete)
-                        echo(CLI_CONFIG, f"done. [{ttot.seconds}s]")
-
-        # Update tracked state information
-        state_data["tracked_backups"] = tracked_backups
-        with open(autobackup_state_file, "w") as fh:
-            jdump(state_data, fh)
-
-        backup_summary[vm] = tracked_backups
-
-    if autobackup_config["auto_mount_enabled"]:
-        # Execute each unmount_cmds command in sequence
-        for cmd in autobackup_config["unmount_cmds"]:
-            echo(
-                CLI_CONFIG,
-                f"Executing unmount command '{cmd.split()[0]}'... ",
-                newline=False,
-            )
-            tstart = datetime.now()
-            ret = run(
-                cmd.split(),
-                stdout=PIPE,
-                stderr=PIPE,
-            )
-            tend = datetime.now()
-            ttot = tend - tstart
-            if ret.returncode != 0:
-                echo(
-                    CLI_CONFIG,
-                    f"failed. [{ttot.seconds}s]",
-                )
-                echo(
-                    CLI_CONFIG,
-                    f"Continuing; command reports: {ret.stderr.decode().strip()}",
-                )
-            else:
-                echo(CLI_CONFIG, f"done. [{ttot.seconds}s]")
-
-    autobackup_end_time = datetime.now()
-    autobackup_total_time = autobackup_end_time - autobackup_start_time
-
-    # Handle report emailing
-    if email_report is not None:
-        echo(CLI_CONFIG, "")
-        echo(CLI_CONFIG, f"Sending email summary report to {email_report}")
-
-        current_datetime = datetime.now()
-        email_datetime = formatdate(float(current_datetime.strftime("%s")))
-
-        email = list()
-        email.append(f"Date: {email_datetime}")
-        email.append(f"Subject: PVC Autobackup report for cluster {cluster}")
-
-        recipients = list()
-        for recipient in email_report.split(","):
-            recipients.append(f"<{recipient}>")
-        email.append(f"To: {', '.join(recipients)}")
-        email.append(f"From: PVC Autobackup System <pvc@{gethostname()}>")
-        email.append("")
-
-        email.append(
-            f"A PVC autobackup has been completed at {current_datetime} in {autobackup_total_time}."
-        )
-        email.append("")
-        email.append(
-            "The following is a summary of all current VM backups after cleanups, most recent first:"
-        )
-        email.append("")
-
-        for vm in backup_vms:
-            email.append(f"VM {vm}:")
-            for backup in backup_summary[vm]:
-                datestring = backup.get("datestring")
-                backup_date = datetime.strptime(datestring, "%Y%m%d%H%M%S")
-                if backup.get("result", False):
-                    email.append(
-                        f"    {backup_date}: Success in {backup.get('runtime_secs', 0)} seconds, ID {datestring}, type {backup.get('type', 'unknown')}"
-                    )
-                    email.append(
-                        f"                         Backup contains {len(backup.get('backup_files'))} files totaling {pvc.lib.storage.format_bytes_tohuman(backup.get('backup_size_bytes', 0))} ({backup.get('backup_size_bytes', 0)} bytes)"
-                    )
-                else:
-                    email.append(
-                        f"    {backup_date}: Failure in {backup.get('runtime_secs', 0)} seconds, ID {datestring}, type {backup.get('type', 'unknown')}"
-                    )
-                    email.append(
-                        f"                         {backup.get('result_message')}"
-                    )
-
-        try:
-            p = popen("/usr/sbin/sendmail -t", "w")
-            p.write("\n".join(email))
-            p.close()
-        except Exception as e:
-            echo(CLI_CONFIG, f"Failed to send report email: {e}")
-
-    echo(CLI_CONFIG, "")
-    echo(CLI_CONFIG, f"Autobackup completed in {autobackup_total_time}.")

client-cli/pvc/lib/vm.py

@@ -595,6 +595,24 @@ def vm_import_snapshot(
     return get_wait_retdata(response, wait_flag)
 
 
+def vm_autobackup(config, email_recipients=None, force_full_flag=False, wait_flag=True):
+    """
+    Perform a cluster VM autobackup
+
+    API endpoint: POST /vm//autobackup
+    API arguments: email_recipients=email_recipients, force_full_flag=force_full_flag
+    API schema: {"message":"{data}"}
+    """
+    params = {
+        "email_recipients": email_recipients,
+        "force_full": force_full_flag,
+    }
+
+    response = call_api(config, "post", "/vm/autobackup", params=params)
+
+    return get_wait_retdata(response, wait_flag)
+
+
 def vm_vcpus_set(config, vm, vcpus, topology, restart):
     """
     Set the vCPU count of the VM with topology

daemon-common/autobackup.py

@@ -0,0 +1,695 @@
+#!/usr/bin/env python3
+
+# autobackup.py - PVC API Autobackup functions
+# Part of the Parallel Virtual Cluster (PVC) system
+#
+#    Copyright (C) 2018-2024 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/>.
+#
+###############################################################################
+
+from datetime import datetime
+from json import load as jload
+from json import dump as jdump
+from os import popen, makedirs, path, scandir
+from shutil import rmtree
+from subprocess import run, PIPE
+
+from daemon_lib.common import run_os_command
+from daemon_lib.config import get_autobackup_configuration
+from daemon_lib.celery import start, fail, log_info, log_err, update, finish
+
+import daemon_lib.ceph as ceph
+import daemon_lib.vm as vm
+
+
+def send_execution_failure_report(
+    celery_conf, config, recipients=None, total_time=0, error=None
+):
+    if recipients is None:
+        return
+
+    from email.utils import formatdate
+    from socket import gethostname
+
+    log_message = f"Sending email failure report to {', '.join(recipients)}"
+    log_info(celery_conf[0], log_message)
+    update(
+        celery_conf[0],
+        log_message,
+        current=celery_conf[1] + 1,
+        total=celery_conf[2],
+    )
+
+    current_datetime = datetime.now()
+    email_datetime = formatdate(float(current_datetime.strftime("%s")))
+
+    email = list()
+    email.append(f"Date: {email_datetime}")
+    email.append(
+        f"Subject: PVC Autobackup execution failure for cluster '{config['cluster']}'"
+    )
+
+    email_to = list()
+    for recipient in recipients:
+        email_to.append(f"<{recipient}>")
+
+    email.append(f"To: {', '.join(email_to)}")
+    email.append(f"From: PVC Autobackup System <pvc@{gethostname()}>")
+    email.append("")
+
+    email.append(
+        f"A PVC autobackup has FAILED at {current_datetime} in {total_time}s due to an execution error."
+    )
+    email.append("")
+    email.append("The reported error message is:")
+    email.append(f"  {error}")
+
+    try:
+        with popen("/usr/sbin/sendmail -t", "w") as p:
+            p.write("\n".join(email))
+    except Exception as e:
+        log_err(f"Failed to send report email: {e}")
+
+
+def send_execution_summary_report(
+    celery_conf, config, recipients=None, total_time=0, summary=dict()
+):
+    if recipients is None:
+        return
+
+    from email.utils import formatdate
+    from socket import gethostname
+
+    log_message = f"Sending email summary report to {', '.join(recipients)}"
+    log_info(celery_conf[0], log_message)
+    update(
+        celery_conf[0],
+        log_message,
+        current=celery_conf[1] + 1,
+        total=celery_conf[2],
+    )
+
+    current_datetime = datetime.now()
+    email_datetime = formatdate(float(current_datetime.strftime("%s")))
+
+    email = list()
+    email.append(f"Date: {email_datetime}")
+    email.append(f"Subject: PVC Autobackup report for cluster '{config['cluster']}'")
+
+    email_to = list()
+    for recipient in recipients:
+        email_to.append(f"<{recipient}>")
+
+    email.append(f"To: {', '.join(email_to)}")
+    email.append(f"From: PVC Autobackup System <pvc@{gethostname()}>")
+    email.append("")
+
+    email.append(
+        f"A PVC autobackup has been completed at {current_datetime} in {total_time}."
+    )
+    email.append("")
+    email.append(
+        "The following is a summary of all current VM backups after cleanups, most recent first:"
+    )
+    email.append("")
+
+    for vm_name in summary.keys():
+        email.append(f"VM: {vm_name}:")
+        for backup in summary[vm_name]:
+            datestring = backup.get("datestring")
+            backup_date = datetime.strptime(datestring, "%Y%m%d%H%M%S")
+            if backup.get("result", False):
+                email.append(
+                    f"    {backup_date}: Success in {backup.get('runtime_secs', 0)} seconds, ID {backup.get('snapshot_name')}, type {backup.get('type', 'unknown')}"
+                )
+                email.append(
+                    f"                         Backup contains {len(backup.get('export_files'))} files totaling {ceph.format_bytes_tohuman(backup.get('export_size_bytes', 0))} ({backup.get('export_size_bytes', 0)} bytes)"
+                )
+            else:
+                email.append(
+                    f"    {backup_date}: Failure in {backup.get('runtime_secs', 0)} seconds, ID {backup.get('snapshot_name')}, type {backup.get('type', 'unknown')}"
+                )
+                email.append(f"                         {backup.get('result_message')}")
+
+    try:
+        with popen("/usr/sbin/sendmail -t", "w") as p:
+            p.write("\n".join(email))
+    except Exception as e:
+        log_err(f"Failed to send report email: {e}")
+
+
+def run_vm_backup(zkhandler, celery, config, vm_detail, force_full=False):
+    vm_name = vm_detail["name"]
+    dom_uuid = vm_detail["uuid"]
+    backup_suffixed_path = f"{config['backup_root_path']}{config['backup_root_suffix']}"
+    vm_backup_path = f"{backup_suffixed_path}/{vm_name}"
+    autobackup_state_file = f"{vm_backup_path}/.autobackup.json"
+    full_interval = config["backup_schedule"]["full_interval"]
+    full_retention = config["backup_schedule"]["full_retention"]
+
+    if not path.exists(vm_backup_path) or not path.exists(autobackup_state_file):
+        # There are no existing backups so the list is empty
+        state_data = dict()
+        tracked_backups = list()
+    else:
+        with open(autobackup_state_file) as fh:
+            state_data = jload(fh)
+        tracked_backups = state_data["tracked_backups"]
+
+    full_backups = [b for b in tracked_backups if b["type"] == "full"]
+    if len(full_backups) > 0:
+        last_full_backup = full_backups[0]
+        last_full_backup_idx = tracked_backups.index(last_full_backup)
+        if force_full:
+            this_backup_incremental_parent = None
+            this_backup_retain_snapshot = True
+        elif last_full_backup_idx >= full_interval - 1:
+            this_backup_incremental_parent = None
+            this_backup_retain_snapshot = True
+        else:
+            this_backup_incremental_parent = last_full_backup["snapshot_name"]
+            this_backup_retain_snapshot = False
+    else:
+        # The very first ackup must be full to start the tree
+        this_backup_incremental_parent = None
+        this_backup_retain_snapshot = True
+
+    export_type = (
+        "incremental" if this_backup_incremental_parent is not None else "full"
+    )
+
+    now = datetime.now()
+    datestring = now.strftime("%Y%m%d%H%M%S")
+    snapshot_name = f"ab{datestring}"
+
+    # Take the VM snapshot (vm.vm_worker_create_snapshot)
+    snap_list = list()
+
+    failure = False
+    export_files = None
+    export_files_size = 0
+
+    def update_tracked_backups():
+        # Read export file to get details
+        backup_json_file = (
+            f"{backup_suffixed_path}/{vm_name}/{snapshot_name}/snapshot.json"
+        )
+        try:
+            with open(backup_json_file) as fh:
+                backup_json = jload(fh)
+            tracked_backups.insert(0, backup_json)
+        except Exception as e:
+            log_err(celery, f"Could not open export JSON: {e}")
+            return list()
+
+        state_data["tracked_backups"] = tracked_backups
+        with open(autobackup_state_file, "w") as fh:
+            jdump(state_data, fh)
+
+        return tracked_backups
+
+    def write_backup_summary(success=False, message=""):
+        ttotal = (datetime.now() - now).total_seconds()
+        export_details = {
+            "type": export_type,
+            "result": success,
+            "message": message,
+            "datestring": datestring,
+            "runtime_secs": ttotal,
+            "snapshot_name": snapshot_name,
+            "incremental_parent": this_backup_incremental_parent,
+            "vm_detail": vm_detail,
+            "export_files": export_files,
+            "export_size_bytes": export_files_size,
+        }
+        try:
+            with open(
+                f"{backup_suffixed_path}/{vm_name}/{snapshot_name}/snapshot.json",
+                "w",
+            ) as fh:
+                jdump(export_details, fh)
+        except Exception as e:
+            log_err(celery, f"Error exporting snapshot details: {e}")
+            return False, e
+
+        return True, ""
+
+    def cleanup_failure():
+        for snapshot in snap_list:
+            rbd, snapshot_name = snapshot.split("@")
+            pool, volume = rbd.split("/")
+            # We capture no output here, because if this fails too we're in a deep
+            # error chain and will just ignore it
+            ceph.remove_snapshot(zkhandler, pool, volume, snapshot_name)
+
+    rbd_list = zkhandler.read(("domain.storage.volumes", dom_uuid)).split(",")
+
+    for rbd in rbd_list:
+        pool, volume = rbd.split("/")
+        ret, msg = ceph.add_snapshot(
+            zkhandler, pool, volume, snapshot_name, zk_only=False
+        )
+        if not ret:
+            cleanup_failure()
+            error_message = msg.replace("ERROR: ", "")
+            log_err(celery, error_message)
+            failure = True
+            break
+        else:
+            snap_list.append(f"{pool}/{volume}@{snapshot_name}")
+
+    if failure:
+        error_message = (f"[{vm_name}] Error in snapshot export, skipping",)
+        write_backup_summary(message=error_message)
+        tracked_backups = update_tracked_backups()
+        return tracked_backups
+
+    # Get the current domain XML
+    vm_config = zkhandler.read(("domain.xml", dom_uuid))
+
+    # Add the snapshot entry to Zookeeper
+    ret = zkhandler.write(
+        [
+            (
+                (
+                    "domain.snapshots",
+                    dom_uuid,
+                    "domain_snapshot.name",
+                    snapshot_name,
+                ),
+                snapshot_name,
+            ),
+            (
+                (
+                    "domain.snapshots",
+                    dom_uuid,
+                    "domain_snapshot.timestamp",
+                    snapshot_name,
+                ),
+                now.strftime("%s"),
+            ),
+            (
+                (
+                    "domain.snapshots",
+                    dom_uuid,
+                    "domain_snapshot.xml",
+                    snapshot_name,
+                ),
+                vm_config,
+            ),
+            (
+                (
+                    "domain.snapshots",
+                    dom_uuid,
+                    "domain_snapshot.rbd_snapshots",
+                    snapshot_name,
+                ),
+                ",".join(snap_list),
+            ),
+        ]
+    )
+    if not ret:
+        error_message = (f"[{vm_name}] Error in snapshot export, skipping",)
+        log_err(celery, error_message)
+        write_backup_summary(message=error_message)
+        tracked_backups = update_tracked_backups()
+        return tracked_backups
+
+    # Export the snapshot (vm.vm_worker_export_snapshot)
+    export_target_path = f"{backup_suffixed_path}/{vm_name}/{snapshot_name}/images"
+
+    try:
+        makedirs(export_target_path)
+    except Exception as e:
+        error_message = (
+            f"[{vm_name}] Failed to create target directory '{export_target_path}': {e}",
+        )
+        log_err(celery, error_message)
+        return tracked_backups
+
+    def export_cleanup():
+        from shutil import rmtree
+
+        rmtree(f"{backup_suffixed_path}/{vm_name}/{snapshot_name}")
+
+    # Set the export filetype
+    if this_backup_incremental_parent is not None:
+        export_fileext = "rbddiff"
+    else:
+        export_fileext = "rbdimg"
+
+    snapshot_volumes = list()
+    for rbdsnap in snap_list:
+        pool, _volume = rbdsnap.split("/")
+        volume, name = _volume.split("@")
+        ret, snapshots = ceph.get_list_snapshot(
+            zkhandler, pool, volume, limit=name, is_fuzzy=False
+        )
+        if ret:
+            snapshot_volumes += snapshots
+
+    export_files = list()
+    for snapshot_volume in snapshot_volumes:
+        snap_pool = snapshot_volume["pool"]
+        snap_volume = snapshot_volume["volume"]
+        snap_snapshot_name = snapshot_volume["snapshot"]
+        snap_size = snapshot_volume["stats"]["size"]
+
+        if this_backup_incremental_parent is not None:
+            retcode, stdout, stderr = run_os_command(
+                f"rbd export-diff --from-snap {this_backup_incremental_parent} {snap_pool}/{snap_volume}@{snap_snapshot_name} {export_target_path}/{snap_pool}.{snap_volume}.{export_fileext}"
+            )
+            if retcode:
+                error_message = (
+                    f"[{vm_name}] Failed to export snapshot for volume(s) '{snap_pool}/{snap_volume}'",
+                )
+                failure = True
+                break
+            else:
+                export_files.append(
+                    (
+                        f"images/{snap_pool}.{snap_volume}.{export_fileext}",
+                        snap_size,
+                    )
+                )
+        else:
+            retcode, stdout, stderr = run_os_command(
+                f"rbd export --export-format 2 {snap_pool}/{snap_volume}@{snap_snapshot_name} {export_target_path}/{snap_pool}.{snap_volume}.{export_fileext}"
+            )
+            if retcode:
+                error_message = (
+                    f"[{vm_name}] Failed to export snapshot for volume(s) '{snap_pool}/{snap_volume}'",
+                )
+                failure = True
+                break
+            else:
+                export_files.append(
+                    (
+                        f"images/{snap_pool}.{snap_volume}.{export_fileext}",
+                        snap_size,
+                    )
+                )
+
+    if failure:
+        log_err(celery, error_message)
+        write_backup_summary(message=error_message)
+        tracked_backups = update_tracked_backups()
+        return tracked_backups
+
+    def get_dir_size(pathname):
+        total = 0
+        with scandir(pathname) as it:
+            for entry in it:
+                if entry.is_file():
+                    total += entry.stat().st_size
+                elif entry.is_dir():
+                    total += get_dir_size(entry.path)
+        return total
+
+    export_files_size = get_dir_size(export_target_path)
+
+    ret, e = write_backup_summary(success=True)
+    if not ret:
+        error_message = (f"[{vm_name}] Failed to export configuration snapshot: {e}",)
+        log_err(celery, error_message)
+        write_backup_summary(message=error_message)
+        tracked_backups = update_tracked_backups()
+        return tracked_backups
+
+    # Clean up the snapshot (vm.vm_worker_remove_snapshot)
+    if not this_backup_retain_snapshot:
+        for snap in snap_list:
+            rbd, name = snap.split("@")
+            pool, volume = rbd.split("/")
+            ret, msg = ceph.remove_snapshot(zkhandler, pool, volume, name)
+            if not ret:
+                error_message = msg.replace("ERROR: ", f"[{vm_name}] ")
+                failure = True
+                break
+
+        if failure:
+            log_err(celery, error_message)
+            write_backup_summary(message=error_message)
+            tracked_backups = update_tracked_backups()
+            return tracked_backups
+
+        ret = zkhandler.delete(
+            ("domain.snapshots", dom_uuid, "domain_snapshot.name", snapshot_name)
+        )
+        if not ret:
+            error_message = (f"[{vm_name}] Failed to remove VM snapshot; continuing",)
+            log_err(celery, error_message)
+
+    marked_for_deletion = list()
+    # Find any full backups that are expired
+    found_full_count = 0
+    for backup in tracked_backups:
+        if backup["type"] == "full":
+            found_full_count += 1
+            if found_full_count > full_retention:
+                marked_for_deletion.append(backup)
+    # Find any incremental backups that depend on marked parents
+    for backup in tracked_backups:
+        if backup["type"] == "incremental" and backup["incremental_parent"] in [
+            b["snapshot_name"] for b in marked_for_deletion
+        ]:
+            marked_for_deletion.append(backup)
+
+    if len(marked_for_deletion) > 0:
+        for backup_to_delete in marked_for_deletion:
+            ret = vm.vm_worker_remove_snapshot(
+                zkhandler, None, vm_name, backup_to_delete["snapshot_name"]
+            )
+            if ret is False:
+                error_message = f"Failed to remove obsolete backup snapshot '{backup_to_delete['snapshot_name']}', leaving in tracked backups"
+                log_err(celery, error_message)
+            else:
+                rmtree(f"{vm_backup_path}/{backup_to_delete['snapshot_name']}")
+                tracked_backups.remove(backup_to_delete)
+
+    tracked_backups = update_tracked_backups()
+    return tracked_backups
+
+
+def worker_cluster_autobackup(
+    zkhandler, celery, force_full=False, email_recipients=None
+):
+    config = get_autobackup_configuration()
+
+    backup_summary = dict()
+
+    current_stage = 0
+    total_stages = 1
+    if email_recipients is not None:
+        total_stages += 1
+
+    start(
+        celery,
+        f"Starting cluster '{config['cluster']}' VM autobackup",
+        current=current_stage,
+        total=total_stages,
+    )
+
+    if not config["autobackup_enabled"]:
+        message = "Autobackups are not configured on this cluster."
+        log_info(celery, message)
+        return finish(
+            celery,
+            message,
+            current=total_stages,
+            total=total_stages,
+        )
+
+    autobackup_start_time = datetime.now()
+
+    retcode, vm_list = vm.get_list(zkhandler)
+    if not retcode:
+        error_message = f"Failed to fetch VM list: {vm_list}"
+        log_err(celery, error_message)
+        send_execution_failure_report(
+            (celery, current_stage, total_stages),
+            config,
+            recipients=email_recipients,
+            error=error_message,
+        )
+        fail(celery, error_message)
+        return False
+
+    backup_suffixed_path = f"{config['backup_root_path']}{config['backup_root_suffix']}"
+    if not path.exists(backup_suffixed_path):
+        makedirs(backup_suffixed_path)
+
+    full_interval = config["backup_schedule"]["full_interval"]
+
+    backup_vms = list()
+    for vm_detail in vm_list:
+        vm_tag_names = [t["name"] for t in vm_detail["tags"]]
+        matching_tags = (
+            True
+            if len(set(vm_tag_names).intersection(set(config["backup_tags"]))) > 0
+            else False
+        )
+        if matching_tags:
+            backup_vms.append(vm_detail)
+
+    if len(backup_vms) < 1:
+        message = "Found no VMs tagged for autobackup."
+        log_info(celery, message)
+        return finish(
+            celery,
+            message,
+            current=total_stages,
+            total=total_stages,
+        )
+
+    if config["auto_mount_enabled"]:
+        total_stages += len(config["mount_cmds"])
+        total_stages += len(config["unmount_cmds"])
+
+    total_stages += len(backup_vms)
+
+    log_info(
+        celery,
+        f"Found {len(backup_vms)} suitable VM(s) for autobackup: {', '.join([b['name'] for b in backup_vms])}",
+    )
+
+    # Handle automount mount commands
+    if config["auto_mount_enabled"]:
+        for cmd in config["mount_cmds"]:
+            current_stage += 1
+            update(
+                celery,
+                f"Executing mount command '{cmd.split()[0]}'",
+                current=current_stage,
+                total=total_stages,
+            )
+
+            ret = run(
+                cmd.split(),
+                stdout=PIPE,
+                stderr=PIPE,
+            )
+
+            if ret.returncode != 0:
+                error_message = f"Failed to execute mount command '{cmd.split()[0]}': {ret.stderr.decode().strip()}"
+                log_err(celery, error_message)
+                send_execution_failure_report(
+                    (celery, current_stage, total_stages),
+                    config,
+                    recipients=email_recipients,
+                    total_time=datetime.now() - autobackup_start_time,
+                    error=error_message,
+                )
+                fail(celery, error_message)
+                return False
+
+    # Execute the backup: take a snapshot, then export the snapshot
+    for vm_detail in backup_vms:
+        vm_backup_path = f"{backup_suffixed_path}/{vm_detail['name']}"
+        autobackup_state_file = f"{vm_backup_path}/.autobackup.json"
+        if not path.exists(vm_backup_path) or not path.exists(autobackup_state_file):
+            # There are no existing backups so the list is empty
+            state_data = dict()
+            tracked_backups = list()
+        else:
+            with open(autobackup_state_file) as fh:
+                state_data = jload(fh)
+            tracked_backups = state_data["tracked_backups"]
+
+        full_backups = [b for b in tracked_backups if b["type"] == "full"]
+        if len(full_backups) > 0:
+            last_full_backup = full_backups[0]
+            last_full_backup_idx = tracked_backups.index(last_full_backup)
+            if force_full:
+                this_backup_incremental_parent = None
+            elif last_full_backup_idx >= full_interval - 1:
+                this_backup_incremental_parent = None
+            else:
+                this_backup_incremental_parent = last_full_backup["snapshot_name"]
+        else:
+            # The very first ackup must be full to start the tree
+            this_backup_incremental_parent = None
+
+        export_type = (
+            "incremental" if this_backup_incremental_parent is not None else "full"
+        )
+
+        current_stage += 1
+        update(
+            celery,
+            f"Performing autobackup of VM {vm_detail['name']} ({export_type})",
+            current=current_stage,
+            total=total_stages,
+        )
+
+        summary = run_vm_backup(
+            zkhandler,
+            celery,
+            config,
+            vm_detail,
+            force_full=force_full,
+        )
+        backup_summary[vm_detail["name"]] = summary
+
+    # Handle automount unmount commands
+    if config["auto_mount_enabled"]:
+        for cmd in config["unmount_cmds"]:
+            current_stage += 1
+            update(
+                celery,
+                f"Executing unmount command '{cmd.split()[0]}'",
+                current=current_stage,
+                total=total_stages,
+            )
+
+            ret = run(
+                cmd.split(),
+                stdout=PIPE,
+                stderr=PIPE,
+            )
+
+            if ret.returncode != 0:
+                error_message = f"Failed to execute unmount command '{cmd.split()[0]}': {ret.stderr.decode().strip()}"
+                log_err(celery, error_message)
+                send_execution_failure_report(
+                    (celery, current_stage, total_stages),
+                    config,
+                    recipients=email_recipients,
+                    total_time=datetime.now() - autobackup_start_time,
+                    error=error_message,
+                )
+                fail(celery, error_message)
+                return False
+
+    autobackup_end_time = datetime.now()
+    autobackup_total_time = autobackup_end_time - autobackup_start_time
+
+    if email_recipients is not None:
+        send_execution_summary_report(
+            (celery, current_stage, total_stages),
+            config,
+            recipients=email_recipients,
+            total_time=autobackup_total_time,
+            summary=backup_summary,
+        )
+        current_stage += 1
+
+    current_stage += 1
+    return finish(
+        celery,
+        f"Successfully completed cluster '{config['cluster']}' VM autobackup",
+        current=current_stage,
+        total=total_stages,
+    )

daemon-common/config.py

@@ -406,6 +406,78 @@ def get_configuration():
     return config
 
 
+def get_parsed_autobackup_configuration(config_file):
+    """
+    Load the configuration; this is the same main pvc.conf that the daemons read
+    """
+    print('Loading configuration from file "{}"'.format(config_file))
+
+    with open(config_file, "r") as cfgfh:
+        try:
+            o_config = yaml.load(cfgfh, Loader=yaml.SafeLoader)
+        except Exception as e:
+            print(f"ERROR: Failed to parse configuration file: {e}")
+            os._exit(1)
+
+    config = dict()
+
+    try:
+        o_cluster = o_config["cluster"]
+        config_cluster = {
+            "cluster": o_cluster["name"],
+            "autobackup_enabled": True,
+        }
+        config = {**config, **config_cluster}
+
+        o_autobackup = o_config["autobackup"]
+        if o_autobackup is None:
+            config["autobackup_enabled"] = False
+            return config
+
+        config_autobackup = {
+            "backup_root_path": o_autobackup["backup_root_path"],
+            "backup_root_suffix": o_autobackup["backup_root_suffix"],
+            "backup_tags": o_autobackup["backup_tags"],
+            "backup_schedule": o_autobackup["backup_schedule"],
+        }
+        config = {**config, **config_autobackup}
+
+        o_automount = o_autobackup["auto_mount"]
+        config_automount = {
+            "auto_mount_enabled": o_automount["enabled"],
+        }
+        config = {**config, **config_automount}
+        if config["auto_mount_enabled"]:
+            config["mount_cmds"] = list()
+            for _mount_cmd in o_automount["mount_cmds"]:
+                if "{backup_root_path}" in _mount_cmd:
+                    _mount_cmd = _mount_cmd.format(
+                        backup_root_path=config["backup_root_path"]
+                    )
+                config["mount_cmds"].append(_mount_cmd)
+            config["unmount_cmds"] = list()
+            for _unmount_cmd in o_automount["unmount_cmds"]:
+                if "{backup_root_path}" in _unmount_cmd:
+                    _unmount_cmd = _unmount_cmd.format(
+                        backup_root_path=config["backup_root_path"]
+                    )
+                config["unmount_cmds"].append(_unmount_cmd)
+
+    except Exception as e:
+        raise MalformedConfigurationError(e)
+
+    return config
+
+
+def get_autobackup_configuration():
+    """
+    Get the configuration.
+    """
+    pvc_config_file = get_configuration_path()
+    config = get_parsed_autobackup_configuration(pvc_config_file)
+    return config
+
+
 def validate_directories(config):
     if not os.path.exists(config["dynamic_directory"]):
         os.makedirs(config["dynamic_directory"])

daemon-common/vm.py

@@ -2074,7 +2074,11 @@ def vm_worker_detach_device(zkhandler, celery, domain, xml_spec):
 
 
 def vm_worker_create_snapshot(
-    zkhandler, celery, domain, snapshot_name=None, zk_only=False
+    zkhandler,
+    celery,
+    domain,
+    snapshot_name=None,
+    zk_only=False,
 ):
     if snapshot_name is None:
         now = datetime.now()
@@ -2117,7 +2121,7 @@ def vm_worker_create_snapshot(
     # Get the list of all RBD volumes
     rbd_list = zkhandler.read(("domain.storage.volumes", dom_uuid)).split(",")
 
-    total_stages = 2 + len(rbd_list)
+    total_stages += 1 + len(rbd_list)
 
     snap_list = list()
 
@@ -2218,7 +2222,12 @@ def vm_worker_create_snapshot(
     )
 
 
-def vm_worker_remove_snapshot(zkhandler, celery, domain, snapshot_name):
+def vm_worker_remove_snapshot(
+    zkhandler,
+    celery,
+    domain,
+    snapshot_name,
+):
     current_stage = 0
     total_stages = 1
     start(
@@ -2275,7 +2284,7 @@ def vm_worker_remove_snapshot(zkhandler, celery, domain, snapshot_name):
     current_stage += 1
     update(
         celery,
-        "Deleting VM configuration snapshot",
+        "Removing VM configuration snapshot",
         current=current_stage,
         total=total_stages,
     )
@@ -2286,7 +2295,7 @@ def vm_worker_remove_snapshot(zkhandler, celery, domain, snapshot_name):
     if not ret:
         fail(
             celery,
-            f'Failed to delete snapshot "{snapshot_name}" of VM "{domain}" in Zookeeper.',
+            f'Failed to remove snapshot "{snapshot_name}" of VM "{domain}" from Zookeeper',
         )
         return False
 
@@ -2398,7 +2407,12 @@ def vm_worker_rollback_snapshot(zkhandler, celery, domain, snapshot_name):
 
 
 def vm_worker_export_snapshot(
-    zkhandler, celery, domain, snapshot_name, export_path, incremental_parent=None
+    zkhandler,
+    celery,
+    domain,
+    snapshot_name,
+    export_path,
+    incremental_parent=None,
 ):
     current_stage = 0
     total_stages = 1
@@ -2422,7 +2436,7 @@ def vm_worker_export_snapshot(
     if not os.path.isdir(export_path):
         fail(
             celery,
-            f"ERROR: Target path '{export_path}' does not exist on node '{myhostname}'",
+            f"Target path '{export_path}' does not exist on node '{myhostname}'",
         )
         return False
 
@@ -2674,9 +2688,7 @@ def vm_worker_import_snapshot(
     # Ensure that the archives are present
     export_source_snapshot_file = f"{vm_import_path}/{snapshot_name}/snapshot.json"
     if not os.path.isfile(export_source_snapshot_file):
-        fail(
-            celery, f"ERROR: The specified source export '{snapshot_name}' do not exist"
-        )
+        fail(celery, f"The specified source export '{snapshot_name}' do not exist")
         return False
 
     # Read the export file and get VM details
@@ -3060,13 +3072,13 @@ def vm_worker_import_snapshot(
             if not retcode:
                 fail(
                     celery,
-                    f"ERROR: Failed to define imported VM: {retmsg}",
+                    f"Failed to define imported VM: {retmsg}",
                 )
                 return False
         except Exception as e:
             fail(
                 celery,
-                f"ERROR: Failed to parse VM export details: {e}",
+                f"Failed to parse VM export details: {e}",
             )
             return False
 
@@ -3095,7 +3107,7 @@ def vm_worker_import_snapshot(
     if not retcode:
         fail(
             celery,
-            f"ERROR: Failed to start imported VM {domain}: {retmsg}",
+            f"Failed to start imported VM {domain}: {retmsg}",
         )
         return False
 

worker-daemon/pvcworkerd/Daemon.py

@@ -47,6 +47,9 @@ from daemon_lib.benchmark import (
 from daemon_lib.vmbuilder import (
     worker_create_vm,
 )
+from daemon_lib.autobackup import (
+    worker_cluster_autobackup,
+)
 
 # Daemon version
 version = "0.9.98"
@@ -101,6 +104,21 @@ def storage_benchmark(self, pool=None, run_on="primary"):
     return run_storage_benchmark(self, pool)
 
 
+@celery.task(name="cluster.autobackup", bind=True, routing_key="run_on")
+def cluster_autobackup(self, force_full=False, email_recipients=None, run_on="primary"):
+    @ZKConnection(config)
+    def run_cluster_autobackup(
+        zkhandler, self, force_full=False, email_recipients=None
+    ):
+        return worker_cluster_autobackup(
+            zkhandler, self, force_full=force_full, email_recipients=email_recipients
+        )
+
+    return run_cluster_autobackup(
+        self, force_full=force_full, email_recipients=email_recipients
+    )
+
+
 @celery.task(name="vm.flush_locks", bind=True, routing_key="run_on")
 def vm_flush_locks(self, domain=None, force_unlock=False, run_on="primary"):
     @ZKConnection(config)