From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from firstgate.proxmox.com (firstgate.proxmox.com [212.224.123.68]) by lore.proxmox.com (Postfix) with ESMTPS id EA5B61FF15C for ; Fri, 14 Nov 2025 15:42:24 +0100 (CET) Received: from firstgate.proxmox.com (localhost [127.0.0.1]) by firstgate.proxmox.com (Proxmox) with ESMTP id 4498F15B9E; Fri, 14 Nov 2025 15:43:05 +0100 (CET) From: Dominik Csapak To: pve-devel@lists.proxmox.com Date: Fri, 14 Nov 2025 15:42:03 +0100 Message-ID: <20251114144225.3639124-5-d.csapak@proxmox.com> X-Mailer: git-send-email 2.47.3 In-Reply-To: <20251114144225.3639124-1-d.csapak@proxmox.com> References: <20251114144225.3639124-1-d.csapak@proxmox.com> MIME-Version: 1.0 X-SPAM-LEVEL: Spam detection results: 0 AWL 0.030 Adjusted score from AWL reputation of From: address BAYES_00 -1.9 Bayes spam probability is 0 to 1% DMARC_MISSING 0.1 Missing DMARC policy KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record SPF_PASS -0.001 SPF: sender matches SPF record Subject: [pve-devel] [PATCH manager v3 2/3] api: implement node-independent bulk actions X-BeenThere: pve-devel@lists.proxmox.com X-Mailman-Version: 2.1.29 Precedence: list List-Id: Proxmox VE development discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: Proxmox VE development discussion Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: pve-devel-bounces@lists.proxmox.com Sender: "pve-devel" To achieve this, start a worker task and use our generic api client to start the tasks on the relevant nodes. The client always points to 'localhost' so we let the pveproxy worry about the proxying etc. We reuse some logic from the startall/stopall/etc. calls, like getting the ordered guest info list. For that to work, we must convert some of the private subs into proper subs. We also fix handling loading configs from other nodes. In each worker, for each task, we check if the target is in the desired state (e.g. stopped when wanting to start, etc.). If that is the case, start the task and put the UPID in a queue to check. This is done until the parallel count is at 'max_workers', at which point we wait until at least one task is finished before starting the next one. Failures (e.g. task errors or failure to fetch the tasks) are printed, and the vmid is saved and they're collectively printed at the end for convenience. Special handling is required for checking the permissions for suspend: We have to load the config of the VM and find the target state storage. We can then check the privileges for that storage. Further improvements can be: * filters (I'd prefer starting out with front end filters) * failure mode resolution (I'd wait until someone requests that) * token handling (probably not necessary since we do check the permissions upfront for the correct token.) Signed-off-by: Dominik Csapak --- changes from v2: * add retry logic for task status calls * use log_warn more consistently * remove 'statestorage' check for individual guests * rename some api-parameter to what the existing gui expects: - 'max-workers' => 'maxworkers' - 'target-node' => 'target' * use 'expose_credentials' to get the api token/ticket instead of creating always a new one. PVE/API2/Cluster.pm | 7 + PVE/API2/Cluster/BulkAction.pm | 45 ++ PVE/API2/Cluster/BulkAction/Guest.pm | 770 +++++++++++++++++++++++++++ PVE/API2/Cluster/BulkAction/Makefile | 17 + PVE/API2/Cluster/Makefile | 4 +- PVE/API2/Nodes.pm | 24 +- 6 files changed, 855 insertions(+), 12 deletions(-) create mode 100644 PVE/API2/Cluster/BulkAction.pm create mode 100644 PVE/API2/Cluster/BulkAction/Guest.pm create mode 100644 PVE/API2/Cluster/BulkAction/Makefile diff --git a/PVE/API2/Cluster.pm b/PVE/API2/Cluster.pm index 83227c8c..d6003a7d 100644 --- a/PVE/API2/Cluster.pm +++ b/PVE/API2/Cluster.pm @@ -25,6 +25,7 @@ use PVE::API2::ACMEAccount; use PVE::API2::ACMEPlugin; use PVE::API2::Backup; use PVE::API2::Cluster::BackupInfo; +use PVE::API2::Cluster::BulkAction; use PVE::API2::Cluster::Ceph; use PVE::API2::Cluster::Mapping; use PVE::API2::Cluster::Jobs; @@ -103,6 +104,11 @@ __PACKAGE__->register_method({ path => 'mapping', }); +__PACKAGE__->register_method({ + subclass => "PVE::API2::Cluster::BulkAction", + path => 'bulk-action', +}); + if ($have_sdn) { __PACKAGE__->register_method({ subclass => "PVE::API2::Network::SDN", @@ -148,6 +154,7 @@ __PACKAGE__->register_method({ { name => 'acme' }, { name => 'backup' }, { name => 'backup-info' }, + { name => 'bulk-action' }, { name => 'ceph' }, { name => 'config' }, { name => 'firewall' }, diff --git a/PVE/API2/Cluster/BulkAction.pm b/PVE/API2/Cluster/BulkAction.pm new file mode 100644 index 00000000..df650514 --- /dev/null +++ b/PVE/API2/Cluster/BulkAction.pm @@ -0,0 +1,45 @@ +package PVE::API2::Cluster::BulkAction; + +use strict; +use warnings; + +use PVE::API2::Cluster::BulkAction::Guest; + +use base qw(PVE::RESTHandler); + +__PACKAGE__->register_method({ + subclass => "PVE::API2::Cluster::BulkAction::Guest", + path => 'guest', +}); + +__PACKAGE__->register_method({ + name => 'index', + path => '', + method => 'GET', + description => "List resource types.", + permissions => { + user => 'all', + }, + parameters => { + additionalProperties => 0, + properties => {}, + }, + returns => { + type => 'array', + items => { + type => "object", + }, + links => [{ rel => 'child', href => "{name}" }], + }, + code => sub { + my ($param) = @_; + + my $result = [ + { name => 'guest' }, + ]; + + return $result; + }, +}); + +1; diff --git a/PVE/API2/Cluster/BulkAction/Guest.pm b/PVE/API2/Cluster/BulkAction/Guest.pm new file mode 100644 index 00000000..daa9740d --- /dev/null +++ b/PVE/API2/Cluster/BulkAction/Guest.pm @@ -0,0 +1,770 @@ +package PVE::API2::Cluster::BulkAction::Guest; + +use strict; +use warnings; + +use PVE::APIClient::LWP; +use PVE::AccessControl; +use PVE::Cluster; +use PVE::Exception qw(raise raise_perm_exc raise_param_exc); +use PVE::INotify; +use PVE::JSONSchema qw(get_standard_option); +use PVE::RESTEnvironment qw(log_warn); +use PVE::RESTHandler; +use PVE::RPCEnvironment; +use PVE::Storage; +use PVE::Tools qw(); + +use PVE::API2::Nodes; + +use base qw(PVE::RESTHandler); + +__PACKAGE__->register_method({ + name => 'index', + path => '', + method => 'GET', + description => "Bulk action index.", + permissions => { user => 'all' }, + parameters => { + additionalProperties => 0, + properties => {}, + }, + returns => { + type => 'array', + items => { + type => "object", + properties => {}, + }, + links => [{ rel => 'child', href => "{name}" }], + }, + code => sub { + my ($param) = @_; + + return [ + { name => 'start' }, + { name => 'shutdown' }, + { name => 'migrate' }, + { name => 'suspend' }, + ]; + }, +}); + +sub create_client { + my ($request_timeout) = @_; + + my $rpcenv = PVE::RPCEnvironment::get(); + my $authuser = $rpcenv->get_user(); + my $credentials = $rpcenv->get_credentials(); + + my $api_token = $credentials->{api_token}; + my $ticket = $credentials->{ticket}; + my $csrf_token = $credentials->{token}; + + my $node = PVE::INotify::nodename(); + my $fingerprint = PVE::Cluster::get_node_fingerprint($node); + + my $conn_args = { + protocol => 'https', + host => 'localhost', # always call the api locally, let pveproxy handle the proxying + port => 8006, + username => $authuser, + ticket => $ticket, + apitoken => $api_token, + timeout => $request_timeout // 25, # default slightly shorter than the proxy->daemon timeout + cached_fingerprints => { + $fingerprint => 1, + }, + }; + + my $api_client = PVE::APIClient::LWP->new($conn_args->%*); + if (defined($csrf_token)) { + $api_client->update_csrftoken($csrf_token); + } + + return $api_client; +} + +sub make_get_request { + my ($client, $path, $retry_count) = @_; + + $retry_count //= 0; + + my $res = eval { $client->get($path) }; + my $err = $@; + if ($err && $retry_count > 0) { + my $retries = 0; + while ($err && $retries < $retry_count) { + $res = eval { $client->get($path) }; + $err = $@; + $retries++; + sleep 1; + } + } + die $err if $err; + return $res; +} + +# starts and awaits a task for each guest given via $startlist. +# +# takes a vm list in the form of +# { +# 0 => { +# 100 => { .. guest info ..}, +# 101 => { .. guest info ..}, +# }, +# 1 => { +# 102 => { .. guest info ..}, +# 103 => { .. guest info ..}, +# }, +# } +# +# max_workers: how many parallel tasks should be started. +# start_task: a sub that returns eiter a upid or 1 (undef means failure) +# check_task: if start_task returned a upid, will wait for that to finish and +# call check_task with the resulting task status +sub handle_task_foreach_guest { + my ($startlist, $max_workers, $start_task, $check_task) = @_; + + my $api_client = create_client(); + + my $failed = []; + for my $order (sort { $a <=> $b } keys $startlist->%*) { + my $vmlist = $startlist->{$order}; + my $workers = {}; + + for my $vmid (sort { $a <=> $b } keys $vmlist->%*) { + + # wait until at least one slot is free + while (scalar(keys($workers->%*)) >= $max_workers) { + for my $upid (keys($workers->%*)) { + my $worker = $workers->{$upid}; + my $node = $worker->{guest}->{node}; + + my $task = + eval { make_get_request($api_client, "/nodes/$node/tasks/$upid/status", 3) }; + if (my $err = $@) { + push $failed->@*, $worker->{vmid}; + + $check_task->($api_client, $worker->{vmid}, $worker->{guest}, 1, undef); + + delete $workers->{$upid}; + } elsif ($task->{status} ne 'running') { + my $is_error = PVE::Tools::upid_status_is_error($task->{exitstatus}); + push $failed->@*, $worker->{vmid} if $is_error; + + $check_task->( + $api_client, $worker->{vmid}, $worker->{guest}, $is_error, $task, + ); + + delete $workers->{$upid}; + } + } + sleep(1); # How much? + } + + my $guest = $vmlist->{$vmid}; + my $upid = eval { $start_task->($api_client, $vmid, $guest) }; + log_warn("$@") if $@; + + # success but no task necessary + next if defined($upid) && "$upid" eq "1"; + + if (!defined($upid)) { + push $failed->@*, $vmid; + next; + } + + $workers->{$upid} = { + vmid => $vmid, + guest => $guest, + }; + } + + # wait until current order is finished + for my $upid (keys($workers->%*)) { + my $worker = $workers->{$upid}; + my $node = $worker->{guest}->{node}; + + my $task = eval { wait_for_task_finished($api_client, $node, $upid) }; + my $err = $@; + my $is_error = ($err || PVE::Tools::upid_status_is_error($task->{exitstatus})) ? 1 : 0; + push $failed->@*, $worker->{vmid} if $is_error; + + $check_task->($api_client, $worker->{vmid}, $worker->{guest}, $is_error, $task); + + delete $workers->{$upid}; + } + } + + return $failed; +} + +sub get_type_text { + my ($type) = @_; + + if ($type eq 'lxc') { + return 'CT'; + } elsif ($type eq 'qemu') { + return 'VM'; + } else { + die "unknown guest type $type\n"; + } +} + +sub wait_for_task_finished { + my ($client, $node, $upid) = @_; + + while (1) { + my $task = make_get_request($client, "/nodes/$node/tasks/$upid/status", 3); + return $task if $task->{status} ne 'running'; + sleep(1); # How much time? + } +} + +sub check_guest_permissions { + my ($rpcenv, $authuser, $vmlist, $priv_list) = @_; + + if (scalar($vmlist->@*) > 0) { + $rpcenv->check($authuser, "/vms/$_", $priv_list) for $vmlist->@*; + } elsif (!$rpcenv->check($authuser, "/", $priv_list, 1)) { + raise_perm_exc("/, " . join(', ', $priv_list->@*)); + } +} + +sub extract_vmlist { + my ($param) = @_; + + if (my $vmlist = $param->{vms}) { + my $vmlist_string = join(',', $vmlist->@*); + return ($vmlist, $vmlist_string); + } + return ([], undef); +} + +sub print_start_action { + my ($vmlist, $prefix, $suffix) = @_; + + $suffix = defined($suffix) ? " $suffix" : ""; + + if (scalar($vmlist->@*)) { + print "$prefix guests$suffix: " . join(', ', $vmlist->@*) . "\n"; + } else { + print "$prefix all guests$suffix\n"; + } +} + +__PACKAGE__->register_method({ + name => 'start', + path => 'start', + method => 'POST', + description => "Bulk start or resume all guests on the cluster.", + permissions => { + description => "The 'VM.PowerMgmt' permission is required on '/' or on '/vms/' for " + . "each ID passed via the 'vms' parameter.", + user => 'all', + }, + protected => 1, + expose_credentials => 1, + parameters => { + additionalProperties => 0, + properties => { + vms => { + description => "Only consider guests from this list of VMIDs.", + type => 'array', + items => get_standard_option('pve-vmid'), + optional => 1, + }, + timeout => { + description => + "Default start timeout in seconds. Only valid for VMs. (default depends on the guest configuration).", + type => 'integer', + optional => 1, + }, + maxworkers => { + description => "How many parallel tasks at maximum should be started.", + optional => 1, + default => 1, + type => 'integer', + }, + # TODO: + # Failure resolution mode (fail, warn, retry?) + # mode-limits (offline only, suspend only, ?) + # filter (tags, name, ?) + }, + }, + returns => { + type => 'string', + description => "UPID of the worker", + }, + code => sub { + my ($param) = @_; + + my $rpcenv = PVE::RPCEnvironment::get(); + my $authuser = $rpcenv->get_user(); + + my ($vmlist, $vmlist_string) = extract_vmlist($param); + + check_guest_permissions($rpcenv, $authuser, $vmlist, ['VM.PowerMgmt']); + + my $code = sub { + my $startlist = + PVE::API2::Nodes::Nodeinfo::get_start_stop_list(undef, undef, $vmlist_string); + + print_start_action($vmlist, "Starting"); + + my $start_task = sub { + my ($api_client, $vmid, $guest) = @_; + my $node = $guest->{node}; + + my $type = $guest->{type}; + my $type_text = get_type_text($type); + my $operation = 'start'; + my $status = eval { + make_get_request($api_client, "/nodes/$node/$type/$vmid/status/current"); + }; + if (defined($status) && $status->{status} eq 'running') { + if (defined($status->{qmpstatus}) && $status->{qmpstatus} ne 'paused') { + log_warn("Skipping $type_text $vmid, already running.\n"); + return 1; + } else { + $operation = 'resume'; + } + } + + my $params = {}; + if (defined($param->{timeout}) && $operation eq 'start' && $type eq 'qemu') { + $params->{timeout} = $param->{timeout}; + } + + my $url = "/nodes/$node/$type/$vmid/status/$operation"; + print "Starting $type_text $vmid\n"; + return $api_client->post($url, $params); + }; + + my $check_task = sub { + my ($api_client, $vmid, $guest, $is_error, $task) = @_; + my $node = $guest->{node}; + + my $default_delay = 0; + + if (!$is_error) { + my $delay = defined($guest->{up}) ? int($guest->{up}) : $default_delay; + if ($delay > 0) { + print "Waiting for $delay seconds (startup delay)\n" + if $guest->{up}; + for (my $i = 0; $i < $delay; $i++) { + sleep(1); + } + } + } else { + my $err = + defined($task) ? $task->{exitstatus} : "could not query task status"; + my $type_text = get_type_text($guest->{type}); + log_warn("Starting $type_text $vmid failed: $err\n"); + } + }; + + my $max_workers = $param->{maxworkers} // 1; + my $failed = + handle_task_foreach_guest($startlist, $max_workers, $start_task, $check_task); + + if (scalar($failed->@*)) { + die "Some guests failed to start: " . join(', ', $failed->@*) . "\n"; + } + }; + + return $rpcenv->fork_worker('bulkstart', undef, $authuser, $code); + }, +}); + +__PACKAGE__->register_method({ + name => 'shutdown', + path => 'shutdown', + method => 'POST', + description => "Bulk shutdown all guests on the cluster.", + permissions => { + description => "The 'VM.PowerMgmt' permission is required on '/' or on '/vms/' for " + . "each ID passed via the 'vms' parameter.", + user => 'all', + }, + protected => 1, + expose_credentials => 1, + parameters => { + additionalProperties => 0, + properties => { + vms => { + description => "Only consider guests from this list of VMIDs.", + type => 'array', + items => get_standard_option('pve-vmid'), + optional => 1, + }, + timeout => { + description => + "Default shutdown timeout in seconds if none is configured for the guest.", + type => 'integer', + default => 180, + optional => 1, + }, + 'force-stop' => { + description => "Makes sure the Guest stops after the timeout.", + type => 'boolean', + default => 1, + optional => 1, + }, + maxworkers => { + description => "How many parallel tasks at maximum should be started.", + optional => 1, + default => 1, + type => 'integer', + }, + # TODO: + # Failure resolution mode (fail, warn, retry?) + # mode-limits (offline only, suspend only, ?) + # filter (tags, name, ?) + }, + }, + returns => { + type => 'string', + description => "UPID of the worker", + }, + code => sub { + my ($param) = @_; + + my $rpcenv = PVE::RPCEnvironment::get(); + my $authuser = $rpcenv->get_user(); + + my ($vmlist, $vmlist_string) = extract_vmlist($param); + + check_guest_permissions($rpcenv, $authuser, $vmlist, ['VM.PowerMgmt']); + + my $code = sub { + my $startlist = + PVE::API2::Nodes::Nodeinfo::get_start_stop_list(undef, undef, $vmlist_string); + + print_start_action($vmlist, "Shutting down"); + + # reverse order for shutdown + for my $order (keys $startlist->%*) { + my $list = delete $startlist->{$order}; + $order = $order * -1; + $startlist->{$order} = $list; + } + + my $start_task = sub { + my ($api_client, $vmid, $guest) = @_; + my $node = $guest->{node}; + + my $type = $guest->{type}; + my $type_text = get_type_text($type); + + my $status = eval { + make_get_request($api_client, "/nodes/$node/$type/$vmid/status/current"); + }; + if (defined($status) && $status->{status} ne 'running') { + log_warn("Skipping $type_text $vmid, not running.\n"); + return 1; + } + + if ( + defined($status) + && defined($status->{qmpstatus}) + && $status->{qmpstatus} eq 'paused' + && !$param->{'force-stop'} + ) { + log_warn("Skipping $type_text $vmid, resume paused VM before shutdown.\n"); + return 1; + } + + my $timeout = int($guest->{down} // $param->{timeout} // 180); + my $forceStop = $param->{'force-stop'} // 1; + + my $params = { + forceStop => $forceStop, + timeout => $timeout, + }; + + my $url = "/nodes/$node/$type/$vmid/status/shutdown"; + print "Shutting down $type_text $vmid (Timeout = $timeout seconds)\n"; + return $api_client->post($url, $params); + }; + + my $check_task = sub { + my ($api_client, $vmid, $guest, $is_error, $task) = @_; + my $node = $guest->{node}; + if ($is_error) { + my $err = + defined($task) ? $task->{exitstatus} : "could not query task status"; + my $type_text = get_type_text($guest->{type}); + log_warn("Stopping $type_text $vmid failed: $err\n"); + } + }; + + my $max_workers = $param->{maxworkers} // 1; + my $failed = + handle_task_foreach_guest($startlist, $max_workers, $start_task, $check_task); + + if (scalar($failed->@*)) { + die "Some guests failed to shutdown " . join(', ', $failed->@*) . "\n"; + } + }; + + return $rpcenv->fork_worker('bulkshutdown', undef, $authuser, $code); + }, +}); + +__PACKAGE__->register_method({ + name => 'suspend', + path => 'suspend', + method => 'POST', + description => "Bulk suspend all guests on the cluster.", + permissions => { + description => + "The 'VM.PowerMgmt' permission is required on '/' or on '/vms/' for each" + . " ID passed via the 'vms' parameter. Additionally, you need 'VM.Config.Disk' on the" + . " '/vms/{vmid}' path and 'Datastore.AllocateSpace' for the configured state-storage(s)", + user => 'all', + }, + protected => 1, + expose_credentials => 1, + parameters => { + additionalProperties => 0, + properties => { + vms => { + description => "Only consider guests from this list of VMIDs.", + type => 'array', + items => get_standard_option('pve-vmid'), + optional => 1, + }, + statestorage => get_standard_option( + 'pve-storage-id', + { + description => "The storage for the VM state.", + requires => 'to-disk', + optional => 1, + completion => \&PVE::Storage::complete_storage_enabled, + }, + ), + 'to-disk' => { + description => + "If set, suspends the guests to disk. Will be resumed on next start.", + type => 'boolean', + default => 0, + optional => 1, + }, + maxworkers => { + description => "How many parallel tasks at maximum should be started.", + optional => 1, + default => 1, + type => 'integer', + }, + # TODO: + # Failure resolution mode (fail, warn, retry?) + # mode-limits (offline only, suspend only, ?) + # filter (tags, name, ?) + }, + }, + returns => { + type => 'string', + description => "UPID of the worker", + }, + code => sub { + my ($param) = @_; + + my $rpcenv = PVE::RPCEnvironment::get(); + my $authuser = $rpcenv->get_user(); + + my ($vmlist, $vmlist_string) = extract_vmlist($param); + + check_guest_permissions($rpcenv, $authuser, $vmlist, ['VM.PowerMgmt']); + + if ($param->{'to-disk'}) { + check_guest_permissions($rpcenv, $authuser, $vmlist, ['VM.Config.Disk']); + if (my $statestorage = $param->{statestorage}) { + $rpcenv->check($authuser, "/storage/$statestorage", + ['Datastore.AllocateSpace']); + } else { + # storage access check will be done by api call itself later + } + } + + my $code = sub { + my $startlist = + PVE::API2::Nodes::Nodeinfo::get_start_stop_list(undef, undef, $vmlist_string); + + print_start_action($vmlist, "Suspending"); + + # reverse order for suspend + for my $order (keys $startlist->%*) { + my $list = delete $startlist->{$order}; + $order = $order * -1; + $startlist->{$order} = $list; + } + + my $start_task = sub { + my ($api_client, $vmid, $guest) = @_; + my $node = $guest->{node}; + + if ($guest->{type} ne 'qemu') { + log_warn("skipping $vmid, only VMs can be suspended"); + return 1; + } + + my $status = + eval { make_get_request($api_client, "/nodes/$node/qemu/$vmid/status/current") }; + if (defined($status) && $status->{status} ne 'running') { + log_warn("Skipping VM $vmid, not running.\n"); + return 1; + } + + my $params = {}; + $params->{'todisk'} = $param->{'to-disk'} // 0; + $params->{statestorage} = $param->{statestorage} + if $param->{'to-disk'} && defined($param->{statestorage}); + + my $url = "/nodes/$node/qemu/$vmid/status/suspend"; + print "Suspending VM $vmid\n"; + return $api_client->post($url, $params); + }; + + my $check_task = sub { + my ($api_client, $vmid, $guest, $is_error, $task) = @_; + my $node = $guest->{node}; + if ($is_error) { + my $err = + defined($task) ? $task->{exitstatus} : "could not query task status"; + my $type_text = get_type_text($guest->{type}); + log_warn("Stopping $type_text $vmid failed: $err\n"); + } + }; + + my $max_workers = $param->{maxworkers} // 1; + my $failed = + handle_task_foreach_guest($startlist, $max_workers, $start_task, $check_task); + + if (scalar($failed->@*)) { + die "Some guests failed to suspend " . join(', ', $failed->@*) . "\n"; + } + }; + + return $rpcenv->fork_worker('bulksuspend', undef, $authuser, $code); + }, +}); + +__PACKAGE__->register_method({ + name => 'migrate', + path => 'migrate', + method => 'POST', + description => "Bulk migrate all guests on the cluster.", + permissions => { + description => + "The 'VM.Migrate' permission is required on '/' or on '/vms/' for each " + . "ID passed via the 'vms' parameter.", + user => 'all', + }, + protected => 1, + expose_credentials => 1, + parameters => { + additionalProperties => 0, + properties => { + vms => { + description => "Only consider guests from this list of VMIDs.", + type => 'array', + items => get_standard_option('pve-vmid'), + optional => 1, + }, + target => get_standard_option('pve-node', { description => "Target node." }), + online => { + type => 'boolean', + description => "Enable live migration for VMs and restart migration for CTs.", + optional => 1, + }, + "with-local-disks" => { + type => 'boolean', + description => "Enable live storage migration for local disk", + optional => 1, + }, + maxworkers => { + description => "How many parallel tasks at maximum should be started.", + optional => 1, + default => 1, + type => 'integer', + }, + # TODO: + # Failure resolution mode (fail, warn, retry?) + # mode-limits (offline only, suspend only, ?) + # filter (tags, name, ?) + }, + }, + returns => { + type => 'string', + description => "UPID of the worker", + }, + code => sub { + my ($param) = @_; + + my $rpcenv = PVE::RPCEnvironment::get(); + my $authuser = $rpcenv->get_user(); + + my ($vmlist, $vmlist_string) = extract_vmlist($param); + + check_guest_permissions($rpcenv, $authuser, $vmlist, ['VM.Migrate']); + + my $code = sub { + my $list = + PVE::API2::Nodes::Nodeinfo::get_filtered_vmlist(undef, $vmlist_string, 1, 1); + + print_start_action($vmlist, "Migrating", "to $param->{target}"); + + my $start_task = sub { + my ($api_client, $vmid, $guest) = @_; + my $node = $guest->{node}; + + my $type = $guest->{type}; + my $type_text = get_type_text($type); + + if ($node eq $param->{target}) { + log_warn("$type_text $vmid already on $node, skipping.\n"); + return 1; + } + + my $params = { + target => $param->{target}, + }; + + if ($type eq 'lxc') { + $params->{restart} = $param->{online} if defined($param->{online}); + } elsif ($type eq 'qemu') { + $params->{online} = $param->{online} if defined($param->{online}); + $params->{'with-local-disks'} = $param->{'with-local-disks'} + if defined($param->{'with-local-disks'}); + } + + my $url = "/nodes/$node/$type/$vmid/migrate"; + print "Migrating $type_text $vmid\n"; + return $api_client->post($url, $params); + }; + + my $check_task = sub { + my ($api_client, $vmid, $guest, $is_error, $task) = @_; + if ($is_error) { + my $err = + defined($task) ? $task->{exitstatus} : "could not query task status"; + my $type_text = get_type_text($guest->{type}); + log_warn("Migrating $type_text $vmid failed: $err\n"); + } + }; + + my $max_workers = $param->{maxworkers} // 1; + my $failed = + handle_task_foreach_guest({ '0' => $list }, $max_workers, $start_task, $check_task); + + if (scalar($failed->@*)) { + die "Some guests failed to migrate " . join(', ', $failed->@*) . "\n"; + } + }; + + return $rpcenv->fork_worker('bulkmigrate', undef, $authuser, $code); + }, +}); + +1; diff --git a/PVE/API2/Cluster/BulkAction/Makefile b/PVE/API2/Cluster/BulkAction/Makefile new file mode 100644 index 00000000..822c1c15 --- /dev/null +++ b/PVE/API2/Cluster/BulkAction/Makefile @@ -0,0 +1,17 @@ +include ../../../../defines.mk + +# for node independent, cluster-wide applicable, API endpoints +# ensure we do not conflict with files shipped by pve-cluster!! +PERLSOURCE= \ + Guest.pm + +all: + +.PHONY: clean +clean: + rm -rf *~ + +.PHONY: install +install: ${PERLSOURCE} + install -d ${PERLLIBDIR}/PVE/API2/Cluster/BulkAction + install -m 0644 ${PERLSOURCE} ${PERLLIBDIR}/PVE/API2/Cluster/BulkAction diff --git a/PVE/API2/Cluster/Makefile b/PVE/API2/Cluster/Makefile index b109e5cb..6cffe4c9 100644 --- a/PVE/API2/Cluster/Makefile +++ b/PVE/API2/Cluster/Makefile @@ -1,11 +1,13 @@ include ../../../defines.mk -SUBDIRS=Mapping +SUBDIRS=Mapping \ + BulkAction # for node independent, cluster-wide applicable, API endpoints # ensure we do not conflict with files shipped by pve-cluster!! PERLSOURCE= \ BackupInfo.pm \ + BulkAction.pm \ MetricServer.pm \ Mapping.pm \ Notifications.pm \ diff --git a/PVE/API2/Nodes.pm b/PVE/API2/Nodes.pm index d0a78b95..66a006ba 100644 --- a/PVE/API2/Nodes.pm +++ b/PVE/API2/Nodes.pm @@ -1867,7 +1867,7 @@ __PACKAGE__->register_method({ # * vmid whitelist # * guest is a template (default: skip) # * guest is HA manged (default: skip) -my $get_filtered_vmlist = sub { +sub get_filtered_vmlist { my ($nodename, $vmfilter, $templates, $ha_managed) = @_; my $vmlist = PVE::Cluster::get_vmlist(); @@ -1894,28 +1894,29 @@ my $get_filtered_vmlist = sub { die "unknown virtual guest type '$d->{type}'\n"; } - my $conf = $class->load_config($vmid); + my $conf = $class->load_config($vmid, $d->{node}); return if !$templates && $class->is_template($conf); return if !$ha_managed && PVE::HA::Config::vm_is_ha_managed($vmid); $res->{$vmid}->{conf} = $conf; $res->{$vmid}->{type} = $d->{type}; $res->{$vmid}->{class} = $class; + $res->{$vmid}->{node} = $d->{node}; }; warn $@ if $@; } return $res; -}; +} # return all VMs which should get started/stopped on power up/down -my $get_start_stop_list = sub { +sub get_start_stop_list { my ($nodename, $autostart, $vmfilter) = @_; # do not skip HA vms on force or if a specific VMID set is wanted my $include_ha_managed = defined($vmfilter) ? 1 : 0; - my $vmlist = $get_filtered_vmlist->($nodename, $vmfilter, undef, $include_ha_managed); + my $vmlist = get_filtered_vmlist($nodename, $vmfilter, undef, $include_ha_managed); my $resList = {}; foreach my $vmid (keys %$vmlist) { @@ -1928,15 +1929,16 @@ my $get_start_stop_list = sub { $resList->{$order}->{$vmid} = $startup; $resList->{$order}->{$vmid}->{type} = $vmlist->{$vmid}->{type}; + $resList->{$order}->{$vmid}->{node} = $vmlist->{$vmid}->{node}; } return $resList; -}; +} my $remove_locks_on_startup = sub { my ($nodename) = @_; - my $vmlist = &$get_filtered_vmlist($nodename, undef, undef, 1); + my $vmlist = get_filtered_vmlist($nodename, undef, undef, 1); foreach my $vmid (keys %$vmlist) { my $conf = $vmlist->{$vmid}->{conf}; @@ -2028,7 +2030,7 @@ __PACKAGE__->register_method({ warn $@ if $@; my $autostart = $force ? undef : 1; - my $startList = $get_start_stop_list->($nodename, $autostart, $param->{vms}); + my $startList = get_start_stop_list($nodename, $autostart, $param->{vms}); # Note: use numeric sorting with <=> for my $order (sort { $a <=> $b } keys %$startList) { @@ -2174,7 +2176,7 @@ __PACKAGE__->register_method({ $rpcenv->{type} = 'priv'; # to start tasks in background - my $stopList = $get_start_stop_list->($nodename, undef, $param->{vms}); + my $stopList = get_start_stop_list($nodename, undef, $param->{vms}); my $cpuinfo = PVE::ProcFSTools::read_cpuinfo(); my $datacenterconfig = cfs_read_file('datacenter.cfg'); @@ -2303,7 +2305,7 @@ __PACKAGE__->register_method({ $rpcenv->{type} = 'priv'; # to start tasks in background - my $toSuspendList = $get_start_stop_list->($nodename, undef, $param->{vms}); + my $toSuspendList = get_start_stop_list($nodename, undef, $param->{vms}); my $cpuinfo = PVE::ProcFSTools::read_cpuinfo(); my $datacenterconfig = cfs_read_file('datacenter.cfg'); @@ -2508,7 +2510,7 @@ __PACKAGE__->register_method({ my $code = sub { $rpcenv->{type} = 'priv'; # to start tasks in background - my $vmlist = &$get_filtered_vmlist($nodename, $param->{vms}, 1, 1); + my $vmlist = get_filtered_vmlist($nodename, $param->{vms}, 1, 1); if (!scalar(keys %$vmlist)) { warn "no virtual guests matched, nothing to do..\n"; return; -- 2.47.3 _______________________________________________ pve-devel mailing list pve-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel