]>
git.proxmox.com Git - pve-container.git/blob - src/PVE/LXC/Migrate.pm
1 package PVE
::LXC
::Migrate
;
7 use File
::Copy
; # fixme: remove
12 use PVE
::ReplicationConfig
;
13 use PVE
::ReplicationState
;
20 use PVE
::AbstractMigrate
;
21 use base
qw(PVE::AbstractMigrate);
23 # compared against remote end's minimum version
24 our $WS_TUNNEL_VERSION = 2;
27 my ($self, $vmid, $code, @param) = @_;
29 return PVE
::LXC
::Config-
>lock_config($vmid, $code, @param);
33 my ($self, $vmid) = @_;
35 my $online = $self->{opts
}->{online
};
36 my $restart= $self->{opts
}->{restart
};
37 my $remote = $self->{opts
}->{remote
};
39 $self->{storecfg
} = PVE
::Storage
::config
();
42 my $conf = $self->{vmconf
} = PVE
::LXC
::Config-
>load_config($vmid);
44 PVE
::LXC
::Config-
>check_lock($conf);
47 if (PVE
::LXC
::check_running
($vmid)) {
48 die "lxc live migration is currently not implemented\n" if $online;
49 die "running container can only be migrated in restart mode" if !$restart;
52 $self->{was_running
} = $running;
55 PVE
::LXC
::Config-
>foreach_volume_full($conf, { include_unused
=> 1 }, sub {
56 my ($ms, $mountpoint) = @_;
58 my $type = $mountpoint->{type
};
59 # skip dev/bind mps when shared
60 if ($type ne 'volume') {
61 if ($mountpoint->{shared
}) {
64 die "cannot migrate local $type mount point '$ms'\n";
68 my $volid = $mountpoint->{volume
} or die "missing volume for mount point '$ms'\n";
70 my ($storage, $volname) = PVE
::Storage
::parse_volume_id
($volid, 1);
71 die "can't determine assigned storage for mount point '$ms'\n" if !$storage;
73 # check if storage is available on both nodes
74 my $scfg = PVE
::Storage
::storage_check_enabled
($self->{storecfg
}, $storage);
76 my $targetsid = $storage;
78 die "content type 'rootdir' is not available on storage '$storage'\n"
79 if !$scfg->{content
}->{rootdir
};
81 if ($scfg->{shared
} && !$remote) {
82 # PVE::Storage::activate_storage checks this for non-shared storages
83 my $plugin = PVE
::Storage
::Plugin-
>lookup($scfg->{type
});
84 warn "Used shared storage '$storage' is not online on source node!\n"
85 if !$plugin->check_connection($storage, $scfg);
87 # unless in restart mode because we shut the container down
88 die "unable to migrate local mount point '$volid' while CT is running"
89 if $running && !$restart;
91 $targetsid = PVE
::JSONSchema
::map_id
($self->{opts
}->{storagemap
}, $storage);
95 my $target_scfg = PVE
::Storage
::storage_check_enabled
($self->{storecfg
}, $targetsid, $self->{node
});
97 die "$volid: content type 'rootdir' is not available on storage '$targetsid'\n"
98 if !$target_scfg->{content
}->{rootdir
};
101 $storages->{$targetsid} = 1;
104 # todo: test if VM uses local resources
107 # test & establish websocket connection
108 my $bridges = map_bridges
($conf, $self->{opts
}->{bridgemap
}, 1);
110 my $remote = $self->{opts
}->{remote
};
111 my $conn = $remote->{conn
};
114 my ($level, $msg) = @_;
115 $self->log($level, $msg);
118 my $websocket_url = "https://$conn->{host}:$conn->{port}/api2/json/nodes/$self->{node}/lxc/$remote->{vmid}/mtunnelwebsocket";
119 my $url = "/nodes/$self->{node}/lxc/$remote->{vmid}/mtunnel";
121 my $tunnel_params = {
122 url
=> $websocket_url,
125 my $storage_list = join(',', keys %$storages);
126 my $bridge_list = join(',', keys %$bridges);
129 storages
=> $storage_list,
130 bridges
=> $bridge_list,
133 my $tunnel = PVE
::Tunnel
::fork_websocket_tunnel
($conn, $url, $req_params, $tunnel_params, $log);
134 my $min_version = $tunnel->{version
} - $tunnel->{age
};
135 $self->log('info', "local WS tunnel version: $WS_TUNNEL_VERSION");
136 $self->log('info', "remote WS tunnel version: $tunnel->{version}");
137 $self->log('info', "minimum required WS tunnel version: $min_version");
138 die "Remote tunnel endpoint not compatible, upgrade required\n"
139 if $WS_TUNNEL_VERSION < $min_version;
140 die "Remote tunnel endpoint too old, upgrade required\n"
141 if $WS_TUNNEL_VERSION > $tunnel->{version
};
143 $self->log('info', "websocket tunnel started\n");
144 $self->{tunnel
} = $tunnel;
146 # test ssh connection
147 my $cmd = [ @{$self->{rem_ssh
}}, '/bin/true' ];
148 eval { $self->cmd_quiet($cmd); };
149 die "Can't connect to destination address using public key\n" if $@;
152 # in restart mode, we shutdown the container before migrating
153 if ($restart && $running) {
154 my $timeout = $self->{opts
}->{timeout
} // 180;
156 $self->log('info', "shutdown CT $vmid\n");
158 PVE
::LXC
::vm_stop
($vmid, 0, $timeout);
167 my ($self, $vmid) = @_;
169 my $remote = $self->{opts
}->{remote
};
171 $self->log('info', "starting migration of CT $self->{vmid} to node '$self->{node}' ($self->{nodeip})");
173 my $conf = $self->{vmconf
};
174 $conf->{lock} = 'migrate';
175 PVE
::LXC
::Config-
>write_config($vmid, $conf);
177 if ($self->{running
}) {
178 $self->log('info', "container is running - using online migration");
181 $self->{volumes
} = []; # list of already migrated volumes
182 my $volhash = {}; # 'config', 'snapshot' or 'storage' for local volumes
183 my $volhash_errors = {};
185 my $path_to_volid = {};
187 my $log_error = sub {
188 my ($msg, $volid) = @_;
190 $volhash_errors->{$volid} = $msg if !defined($volhash_errors->{$volid});
194 my $test_volid = sub {
195 my ($volid, $snapname) = @_;
199 # check if volume exists, might be a pending new one
200 if ($volid =~ $PVE::LXC
::NEW_DISK_RE
) {
201 $self->log('info', "volume '$volid' does not exist (pending change?)");
205 my ($sid, $volname) = PVE
::Storage
::parse_volume_id
($volid);
207 # check if storage is available on source node
208 my $scfg = PVE
::Storage
::storage_check_enabled
($self->{storecfg
}, $sid);
210 my $targetsid = $sid;
212 if ($scfg->{shared
} && !$remote) {
213 $self->log('info', "volume '$volid' is on shared storage '$sid'")
217 $targetsid = PVE
::JSONSchema
::map_id
($self->{opts
}->{storagemap
}, $sid);
220 PVE
::Storage
::storage_check_enabled
($self->{storecfg
}, $targetsid, $self->{node
})
223 my $bwlimit = $self->get_bwlimit($sid, $targetsid);
225 $volhash->{$volid}->{ref} = defined($snapname) ?
'snapshot' : 'config';
226 $volhash->{$volid}->{snapshots
} = 1 if defined($snapname);
227 $volhash->{$volid}->{targetsid
} = $targetsid;
228 $volhash->{$volid}->{bwlimit
} = $bwlimit;
230 my ($path, $owner) = PVE
::Storage
::path
($self->{storecfg
}, $volid);
232 die "owned by other guest (owner = $owner)\n"
233 if !$owner || ($owner != $self->{vmid
});
235 $path_to_volid->{$path}->{$volid} = 1;
237 if (defined($snapname)) {
238 # we cannot migrate shapshots on local storage
239 # exceptions: 'zfspool', 'btrfs'
240 if ($scfg->{type
} eq 'zfspool' || $scfg->{type
} eq 'btrfs') {
243 die "non-migratable snapshot exists\n";
248 my ($ms, $mountpoint, $snapname) = @_;
250 my $volid = $mountpoint->{volume
};
251 # already checked in prepare
252 if ($mountpoint->{type
} ne 'volume') {
253 $self->log('info', "ignoring shared '$mountpoint->{type}' mount point '$ms' ('$volid')")
259 &$test_volid($volid, $snapname);
261 die "remote migration with snapshots not supported yet\n"
262 if $remote && $snapname;
265 &$log_error($@, $volid) if $@;
268 # first all volumes referenced in snapshots
269 foreach my $snapname (keys %{$conf->{snapshots
}}) {
270 &$test_volid($conf->{snapshots
}->{$snapname}->{'vmstate'}, 0, undef)
271 if defined($conf->{snapshots
}->{$snapname}->{'vmstate'});
272 PVE
::LXC
::Config-
>foreach_volume($conf->{snapshots
}->{$snapname}, $test_mp, $snapname);
275 # then all pending volumes
276 if (defined($conf->{pending
}) && $conf->{pending
}->%*) {
277 PVE
::LXC
::Config-
>foreach_volume($conf->{pending
}, $test_mp);
280 # finally all current volumes
281 PVE
::LXC
::Config-
>foreach_volume_full($conf, { include_unused
=> 1 }, $test_mp);
283 for my $path (keys %$path_to_volid) {
284 my @volids = keys $path_to_volid->{$path}->%*;
285 die "detected not supported aliased volumes: '" . join("', '", @volids) . "'\n"
286 if (scalar @volids > 1);
289 # additional checks for local storage
290 foreach my $volid (keys %$volhash) {
292 my ($sid, $volname) = PVE
::Storage
::parse_volume_id
($volid);
293 my $scfg = PVE
::Storage
::storage_config
($self->{storecfg
}, $sid);
295 # TODO move to storage plugin layer?
296 my $migratable_storages = [
304 push @$migratable_storages, 'cifs';
305 push @$migratable_storages, 'nfs';
308 die "storage type '$scfg->{type}' not supported\n"
309 if !grep { $_ eq $scfg->{type
} } @$migratable_storages;
311 # image is a linked clone on local storage, se we can't migrate.
312 if (my $basename = (PVE
::Storage
::parse_volname
($self->{storecfg
}, $volid))[3]) {
313 die "clone of '$basename'";
316 &$log_error($@, $volid) if $@;
319 foreach my $volid (sort keys %$volhash) {
320 my $ref = $volhash->{$volid}->{ref};
321 if ($ref eq 'storage') {
322 $self->log('info', "found local volume '$volid' (via storage)\n");
323 } elsif ($ref eq 'config') {
324 $self->log('info', "found local volume '$volid' (in current VM config)\n");
325 } elsif ($ref eq 'snapshot') {
326 $self->log('info', "found local volume '$volid' (referenced by snapshot(s))\n");
328 $self->log('info', "found local volume '$volid'\n");
332 foreach my $volid (sort keys %$volhash_errors) {
333 $self->log('warn', "can't migrate local volume '$volid': $volhash_errors->{$volid}");
337 die "can't migrate CT - check log\n";
342 my $rep_cfg = PVE
::ReplicationConfig-
>new();
345 die "cannot remote-migrate replicated VM\n"
346 if $rep_cfg->check_for_existing_jobs($vmid, 1);
347 } elsif (my $jobcfg = $rep_cfg->find_local_replication_job($vmid, $self->{node
})) {
348 die "can't live migrate VM with replicated volumes\n" if $self->{running
};
349 my $start_time = time();
350 my $logfunc = sub { my ($msg) = @_; $self->log('info', $msg); };
351 $rep_volumes = PVE
::Replication
::run_replication
(
352 'PVE::LXC::Config', $jobcfg, $start_time, $start_time, $logfunc);
355 my $opts = $self->{opts
};
356 foreach my $volid (keys %$volhash) {
357 next if $rep_volumes->{$volid};
358 push @{$self->{volumes
}}, $volid;
360 # JSONSchema and get_bandwidth_limit use kbps - storage_migrate bps
361 my $bwlimit = $volhash->{$volid}->{bwlimit
};
362 $bwlimit = $bwlimit * 1024 if defined($bwlimit);
364 my $targetsid = $volhash->{$volid}->{targetsid
};
366 my $new_volid = eval {
369 my ($level, $msg) = @_;
370 $self->log($level, $msg);
373 return PVE
::StorageTunnel
::storage_migrate
(
383 my $storage_migrate_opts = {
384 'ratelimit_bps' => $bwlimit,
385 'insecure' => $opts->{migration_type
} eq 'insecure',
386 'with_snapshots' => $volhash->{$volid}->{snapshots
},
390 my $logfunc = sub { $self->log('info', $_[0]); };
391 return PVE
::Storage
::storage_migrate
(
396 $storage_migrate_opts,
403 die "storage migration for '$volid' to storage '$targetsid' failed - $err\n";
406 $self->{volume_map
}->{$volid} = $new_volid;
407 $self->log('info', "volume '$volid' is '$new_volid' on the target\n");
409 eval { PVE
::Storage
::deactivate_volumes
($self->{storecfg
}, [$volid]); };
411 $self->log('warn', $err);
415 if ($self->{running
}) {
419 # make sure everything on (shared) storage is unmounted
420 # Note: we must be 100% sure, else we get data corruption because
421 # non-shared file system could be mounted twice (on shared storage)
423 PVE
::LXC
::umount_all
($vmid, $self->{storecfg
}, $conf);
425 #to be sure there are no active volumes
426 my $vollist = PVE
::LXC
::Config-
>get_vm_volumes($conf);
427 PVE
::Storage
::deactivate_volumes
($self->{storecfg
}, $vollist);
430 my $remote_conf = PVE
::LXC
::Config-
>load_config($vmid);
431 PVE
::LXC
::Config-
>update_volume_ids($remote_conf, $self->{volume_map
});
433 my $bridges = map_bridges
($remote_conf, $self->{opts
}->{bridgemap
});
434 for my $target (keys $bridges->%*) {
435 for my $nic (keys $bridges->{$target}->%*) {
436 $self->log('info', "mapped: $nic from $bridges->{$target}->{$nic} to $target");
439 my $conf_str = PVE
::LXC
::Config
::write_pct_config
("remote", $remote_conf);
441 # TODO expose in PVE::Firewall?
442 my $vm_fw_conf_path = "/etc/pve/firewall/$vmid.fw";
444 $fw_conf_str = PVE
::Tools
::file_get_contents
($vm_fw_conf_path)
445 if -e
$vm_fw_conf_path;
448 'firewall-config' => $fw_conf_str,
451 PVE
::Tunnel
::write_tunnel
($self->{tunnel
}, 10, 'config', $params);
453 # transfer replication state before moving config
454 $self->transfer_replication_state() if $rep_volumes;
455 PVE
::LXC
::Config-
>update_volume_ids($conf, $self->{volume_map
});
456 PVE
::LXC
::Config-
>write_config($vmid, $conf);
457 PVE
::LXC
::Config-
>move_config_to_node($vmid, $self->{node
});
458 $self->switch_replication_job_target() if $rep_volumes;
460 $self->{conf_migrated
} = 1;
464 my ($self, $vmid, $err) = @_;
466 $self->log('info', "aborting phase 1 - cleanup resources");
468 if ($self->{volumes
}) {
469 foreach my $volid (@{$self->{volumes
}}) {
470 if (my $mapped_volume = $self->{volume_map
}->{$volid}) {
471 $volid = $mapped_volume;
473 $self->log('err', "found stale volume copy '$volid' on node '$self->{node}'");
474 # fixme: try to remove ?
478 if ($self->{opts
}->{remote
}) {
479 # cleans up remote volumes
480 PVE
::Tunnel
::finish_tunnel
($self->{tunnel
}, 1);
481 delete $self->{tunnel
};
486 my ($self, $vmid) = @_;
488 my $volids = $self->{volumes
};
490 # handled below in final_cleanup
491 return if $self->{opts
}->{remote
};
493 # destroy local copies
494 foreach my $volid (@$volids) {
495 eval { PVE
::Storage
::vdisk_free
($self->{storecfg
}, $volid); };
497 $self->log('err', "removing local copy of '$volid' failed - $err");
499 last if $err =~ /^interrupted by signal$/;
505 my ($self, $vmid) = @_;
507 $self->log('info', "start final cleanup");
509 if (!$self->{conf_migrated
}) {
510 eval { PVE
::LXC
::Config-
>remove_lock($vmid, 'migrate'); };
512 $self->log('err', $err);
514 # in restart mode, we start the container on the source node on migration error
515 if ($self->{opts
}->{restart
} && $self->{was_running
}) {
516 $self->log('info', "start container on source node");
518 PVE
::LXC
::vm_start
($vmid, $self->{vmconf
}, $skiplock);
520 } elsif ($self->{opts
}->{remote
}) {
521 eval { PVE
::Tunnel
::write_tunnel
($self->{tunnel
}, 10, 'unlock') };
522 $self->log('err', "Failed to clear migrate lock - $@\n") if $@;
524 if ($self->{opts
}->{restart
} && $self->{was_running
}) {
525 $self->log('info', "start container on target node");
526 PVE
::Tunnel
::write_tunnel
($self->{tunnel
}, 60, 'start');
528 if ($self->{opts
}->{delete}) {
529 PVE
::LXC
::destroy_lxc_container
(
530 PVE
::Storage
::config
(),
532 PVE
::LXC
::Config-
>load_config($vmid),
537 PVE
::Tunnel
::finish_tunnel
($self->{tunnel
});
539 my $cmd = [ @{$self->{rem_ssh
}}, 'pct', 'unlock', $vmid ];
540 $self->cmd_logerr($cmd, errmsg
=> "failed to clear migrate lock");
542 # in restart mode, we start the container on the target node after migration
543 if ($self->{opts
}->{restart
} && $self->{was_running
}) {
544 $self->log('info', "start container on target node");
545 my $cmd = [ @{$self->{rem_ssh
}}, 'pct', 'start', $vmid];
552 my ($conf, $map, $scan_only) = @_;
556 foreach my $opt (keys %$conf) {
557 next if $opt !~ m/^net\d+$/;
559 next if !$conf->{$opt};
560 my $d = PVE
::LXC
::Config-
>parse_lxc_network($conf->{$opt});
561 next if !$d || !$d->{bridge
};
563 my $target_bridge = PVE
::JSONSchema
::map_id
($map, $d->{bridge
});
564 $bridges->{$target_bridge}->{$opt} = $d->{bridge
};
568 $d->{bridge
} = $target_bridge;
569 $conf->{$opt} = PVE
::LXC
::Config-
>print_lxc_network($d);