1 package PVE
::CLI
::pve7to8
;
10 use PVE
::API2
::Certificates
;
11 use PVE
::API2
::Cluster
::Ceph
;
13 use PVE
::AccessControl
;
20 use PVE
::RPCEnvironment
;
22 use PVE
::Storage
::Plugin
;
23 use PVE
::Tools
qw(run_command split_list);
26 use PVE
::VZDump
::Common
;
35 use base
qw(PVE::CLIHandler);
37 my $nodename = PVE
::INotify
::nodename
();
39 sub setup_environment
{
40 PVE
::RPCEnvironment-
>setup_default_cli_env();
43 my ($min_pve_major, $min_pve_minor, $min_pve_pkgrel) = (7, 4, 1);
45 my $forced_legacy_cgroup = 0;
55 my ($level, $line) = @_;
57 $counters->{$level}++ if defined($level) && defined($counters->{$level});
59 print uc($level), ': ' if defined($level);
65 $log_line->('pass', @_);
70 $log_line->('info', @_);
73 $log_line->('skip', @_);
76 print color
('yellow');
77 $log_line->('warn', @_);
82 $log_line->('fail', @_);
86 my $print_header_first = 1;
89 print "\n" if !$print_header_first;
91 $print_header_first = 0;
94 my $get_systemd_unit_state = sub {
98 my $filter_output = sub {
103 run_command
(['systemctl', 'is-enabled', "$unit"], outfunc
=> $filter_output, noerr
=> 1);
104 return if !defined($state);
105 run_command
(['systemctl', 'is-active', "$unit"], outfunc
=> $filter_output, noerr
=> 1);
108 return $state // 'unknown';
110 my $log_systemd_unit_state = sub {
111 my ($unit, $no_fail_on_inactive) = @_;
113 my $log_method = \
&log_warn
;
115 my $state = $get_systemd_unit_state->($unit);
116 if ($state eq 'active') {
117 $log_method = \
&log_pass
;
118 } elsif ($state eq 'inactive') {
119 $log_method = $no_fail_on_inactive ? \
&log_warn
: \
&log_fail
;
120 } elsif ($state eq 'failed') {
121 $log_method = \
&log_fail
;
124 $log_method->("systemd unit '$unit' is in state '$state'");
131 $versions = eval { PVE
::API2
::APT-
>versions({ node
=> $nodename }) } if !defined($versions);
133 if (!defined($versions)) {
134 my $msg = "unable to retrieve package version information";
135 $msg .= "- $@" if $@;
140 my $pkgs = [ grep { $_->{Package
} eq $pkg } @$versions ];
141 if (!defined $pkgs || $pkgs == 0) {
142 log_fail
("unable to determine installed $pkg version.");
149 sub check_pve_packages
{
150 print_header
("CHECKING VERSION INFORMATION FOR PVE PACKAGES");
152 print "Checking for package updates..\n";
153 my $updates = eval { PVE
::API2
::APT-
>list_updates({ node
=> $nodename }); };
154 if (!defined($updates)) {
155 log_warn
("$@") if $@;
156 log_fail
("unable to retrieve list of package updates!");
157 } elsif (@$updates > 0) {
158 my $pkgs = join(', ', map { $_->{Package
} } @$updates);
159 log_warn
("updates for the following packages are available:\n $pkgs");
161 log_pass
("all packages uptodate");
164 print "\nChecking proxmox-ve package version..\n";
165 if (defined(my $proxmox_ve = $get_pkg->('proxmox-ve'))) {
166 my $min_pve_ver = "$min_pve_major.$min_pve_minor-$min_pve_pkgrel";
168 my ($maj, $min, $pkgrel) = $proxmox_ve->{OldVersion
} =~ m/^(\d+)\.(\d+)-(\d+)/;
172 if ($maj > $min_pve_major) {
173 log_pass
("already upgraded to Proxmox VE " . ($min_pve_major + 1));
175 } elsif ($maj >= $min_pve_major && $min >= $min_pve_minor && $pkgrel >= $min_pve_pkgrel) {
176 log_pass
("proxmox-ve package has version >= $min_pve_ver");
178 log_fail
("proxmox-ve package is too old, please upgrade to >= $min_pve_ver!");
181 my ($krunning, $kinstalled) = (qr/6\.(?:2|5)/, 'pve-kernel-6.2');
183 # we got a few that avoided 5.15 in cluster with mixed CPUs, so allow older too
184 ($krunning, $kinstalled) = (qr/(?:5\.(?:13|15)|6\.2)/, 'pve-kernel-5.15');
187 print "\nChecking running kernel version..\n";
188 my $kernel_ver = $proxmox_ve->{RunningKernel
};
189 if (!defined($kernel_ver)) {
190 log_fail
("unable to determine running kernel version.");
191 } elsif ($kernel_ver =~ /^$krunning/) {
192 log_pass
("running kernel '$kernel_ver' is considered suitable for upgrade.");
193 } elsif ($get_pkg->($kinstalled)) {
194 log_warn
("expected kernel '$kinstalled' intalled but not yet rebooted!");
196 log_warn
("unexpected running and installed kernel '$kernel_ver'.");
199 log_fail
("proxmox-ve package not found!");
204 sub check_storage_health
{
205 print_header
("CHECKING CONFIGURED STORAGES");
206 my $cfg = PVE
::Storage
::config
();
210 my $info = PVE
::Storage
::storage_info
($cfg);
212 foreach my $storeid (sort keys %$info) {
213 my $d = $info->{$storeid};
216 log_pass
("storage '$storeid' enabled and active.");
218 log_warn
("storage '$storeid' enabled but not active!");
221 log_skip
("storage '$storeid' disabled.");
225 check_storage_content
();
228 sub check_cluster_corosync
{
229 print_header
("CHECKING CLUSTER HEALTH/SETTINGS");
231 if (!PVE
::Corosync
::check_conf_exists
(1)) {
232 log_skip
("standalone node.");
236 $log_systemd_unit_state->('pve-cluster.service');
237 $log_systemd_unit_state->('corosync.service');
239 if (PVE
::Cluster
::check_cfs_quorum
(1)) {
240 log_pass
("Cluster Filesystem is quorate.");
242 log_fail
("Cluster Filesystem readonly, lost quorum?!");
245 my $conf = PVE
::Cluster
::cfs_read_file
('corosync.conf');
246 my $conf_nodelist = PVE
::Corosync
::nodelist
($conf);
249 print "\nAnalzying quorum settings and state..\n";
250 if (!defined($conf_nodelist)) {
251 log_fail
("unable to retrieve nodelist from corosync.conf");
253 if (grep { $conf_nodelist->{$_}->{quorum_votes
} != 1 } keys %$conf_nodelist) {
254 log_warn
("non-default quorum_votes distribution detected!");
256 map { $node_votes += $conf_nodelist->{$_}->{quorum_votes
} // 0 } keys %$conf_nodelist;
259 my ($expected_votes, $total_votes);
260 my $filter_output = sub {
262 ($expected_votes) = $line =~ /^Expected votes:\s*(\d+)\s*$/
263 if !defined($expected_votes);
264 ($total_votes) = $line =~ /^Total votes:\s*(\d+)\s*$/
265 if !defined($total_votes);
268 run_command
(['corosync-quorumtool', '-s'], outfunc
=> $filter_output, noerr
=> 1);
271 if (!defined($expected_votes)) {
272 log_fail
("unable to get expected number of votes, assuming 0.");
275 if (!defined($total_votes)) {
276 log_fail
("unable to get expected number of votes, assuming 0.");
280 my $cfs_nodelist = PVE
::Cluster
::get_clinfo
()->{nodelist
};
281 my $offline_nodes = grep { $cfs_nodelist->{$_}->{online
} != 1 } keys %$cfs_nodelist;
282 if ($offline_nodes > 0) {
283 log_fail
("$offline_nodes nodes are offline!");
286 my $qdevice_votes = 0;
287 if (my $qdevice_setup = $conf->{main
}->{quorum
}->{device
}) {
288 $qdevice_votes = $qdevice_setup->{votes
} // 1;
291 log_info
("configured votes - nodes: $node_votes");
292 log_info
("configured votes - qdevice: $qdevice_votes");
293 log_info
("current expected votes: $expected_votes");
294 log_info
("current total votes: $total_votes");
296 log_warn
("expected votes set to non-standard value '$expected_votes'.")
297 if $expected_votes != $node_votes + $qdevice_votes;
298 log_warn
("total votes < expected votes: $total_votes/$expected_votes!")
299 if $total_votes < $expected_votes;
301 my $conf_nodelist_count = scalar(keys %$conf_nodelist);
302 my $cfs_nodelist_count = scalar(keys %$cfs_nodelist);
303 log_warn
("cluster consists of less than three quorum-providing nodes!")
304 if $conf_nodelist_count < 3 && $conf_nodelist_count + $qdevice_votes < 3;
306 log_fail
("corosync.conf ($conf_nodelist_count) and pmxcfs ($cfs_nodelist_count) don't agree about size of nodelist.")
307 if $conf_nodelist_count != $cfs_nodelist_count;
309 print "\nChecking nodelist entries..\n";
310 my $nodelist_pass = 1;
311 for my $cs_node (sort keys %$conf_nodelist) {
312 my $entry = $conf_nodelist->{$cs_node};
313 if (!defined($entry->{name
})) {
315 log_fail
("$cs_node: no name entry in corosync.conf.");
317 if (!defined($entry->{nodeid
})) {
319 log_fail
("$cs_node: no nodeid configured in corosync.conf.");
322 for my $link (0..7) {
323 $gotLinks++ if defined($entry->{"ring${link}_addr"});
325 if ($gotLinks <= 0) {
327 log_fail
("$cs_node: no ringX_addr (0 <= X <= 7) link defined in corosync.conf.");
330 my $verify_ring_ip = sub {
332 if (defined(my $ring = $entry->{$key})) {
333 my ($resolved_ip, undef) = PVE
::Corosync
::resolve_hostname_like_corosync
($ring, $conf);
334 if (defined($resolved_ip)) {
335 if ($resolved_ip ne $ring) {
338 "$cs_node: $key '$ring' resolves to '$resolved_ip'.\n"
339 ." Consider replacing it with the currently resolved IP address."
345 "$cs_node: unable to resolve $key '$ring' to an IP address according to Corosync's"
346 ." resolve strategy - cluster will potentially fail with Corosync 3.x/kronosnet!"
351 for my $link (0..7) {
352 $verify_ring_ip->("ring${link}_addr");
355 log_pass
("nodelist settings OK") if $nodelist_pass;
357 print "\nChecking totem settings..\n";
358 my $totem = $conf->{main
}->{totem
};
361 my $transport = $totem->{transport
};
362 if (defined($transport)) {
363 if ($transport ne 'knet') {
365 log_fail
("Corosync transport explicitly set to '$transport' instead of implicit default!");
369 # TODO: are those values still up-to-date?
370 if ((!defined($totem->{secauth
}) || $totem->{secauth
} ne 'on') && (!defined($totem->{crypto_cipher
}) || $totem->{crypto_cipher
} eq 'none')) {
372 log_fail
("Corosync authentication/encryption is not explicitly enabled (secauth / crypto_cipher / crypto_hash)!");
373 } elsif (defined($totem->{crypto_cipher
}) && $totem->{crypto_cipher
} eq '3des') {
375 log_fail
("Corosync encryption cipher set to '3des', no longer supported in Corosync 3.x!"); # FIXME: can be removed?
378 log_pass
("totem settings OK") if $totem_pass;
380 log_info
("run 'pvecm status' to get detailed cluster status..");
382 if (defined(my $corosync = $get_pkg->('corosync'))) {
383 if ($corosync->{OldVersion
} =~ m/^2\./) {
384 log_fail
("\ncorosync 2.x installed, cluster-wide upgrade to 3.x needed!");
385 } elsif ($corosync->{OldVersion
} !~ m/^3\./) {
386 log_fail
("\nunexpected corosync version installed: $corosync->{OldVersion}!");
392 print_header
("CHECKING HYPER-CONVERGED CEPH STATUS");
394 if (PVE
::Ceph
::Tools
::check_ceph_inited
(1)) {
395 log_info
("hyper-converged ceph setup detected!");
397 log_skip
("no hyper-converged ceph setup detected!");
401 log_info
("getting Ceph status/health information..");
402 my $ceph_status = eval { PVE
::API2
::Ceph-
>status({ node
=> $nodename }); };
403 my $noout = eval { PVE
::API2
::Cluster
::Ceph-
>get_flag({ flag
=> "noout" }); };
405 log_fail
("failed to get 'noout' flag status - $@");
408 my $noout_wanted = 1;
410 if (!$ceph_status || !$ceph_status->{health
}) {
411 log_fail
("unable to determine Ceph status!");
413 my $ceph_health = $ceph_status->{health
}->{status
};
415 log_fail
("unable to determine Ceph health!");
416 } elsif ($ceph_health eq 'HEALTH_OK') {
417 log_pass
("Ceph health reported as 'HEALTH_OK'.");
418 } elsif ($ceph_health eq 'HEALTH_WARN' && $noout && (keys %{$ceph_status->{health
}->{checks
}} == 1)) {
419 log_pass
("Ceph health reported as 'HEALTH_WARN' with a single failing check and 'noout' flag set.");
421 log_warn
("Ceph health reported as '$ceph_health'.\n Use the PVE ".
422 "dashboard or 'ceph -s' to determine the specific issues and try to resolve them.");
426 # TODO: check OSD min-required version, if to low it breaks stuff!
428 log_info
("getting Ceph daemon versions..");
429 my $ceph_versions = eval { PVE
::Ceph
::Tools
::get_cluster_versions
(undef, 1); };
430 if (!$ceph_versions) {
431 log_fail
("unable to determine Ceph daemon versions!");
434 { 'key' => 'mon', 'name' => 'monitor' },
435 { 'key' => 'mgr', 'name' => 'manager' },
436 { 'key' => 'mds', 'name' => 'MDS' },
437 { 'key' => 'osd', 'name' => 'OSD' },
440 foreach my $service (@$services) {
441 my $name = $service->{name
};
442 if (my $service_versions = $ceph_versions->{$service->{key
}}) {
443 if (keys %$service_versions == 0) {
444 log_skip
("no running instances detected for daemon type $name.");
445 } elsif (keys %$service_versions == 1) {
446 log_pass
("single running version detected for daemon type $name.");
448 log_warn
("multiple running versions detected for daemon type $name!");
451 log_skip
("unable to determine versions of running Ceph $name instances.");
455 my $overall_versions = $ceph_versions->{overall
};
456 if (!$overall_versions) {
457 log_warn
("unable to determine overall Ceph daemon versions!");
458 } elsif (keys %$overall_versions == 1) {
459 log_pass
("single running overall version detected for all Ceph daemon types.");
460 $noout_wanted = 0; # off post-upgrade, on pre-upgrade
462 log_warn
("overall version mismatch detected, check 'ceph versions' output for details!");
468 log_pass
("'noout' flag set to prevent rebalancing during cluster-wide upgrades.");
470 log_warn
("'noout' flag set, Ceph cluster upgrade seems finished.");
472 } elsif ($noout_wanted) {
473 log_warn
("'noout' flag not set - recommended to prevent rebalancing during upgrades.");
476 log_info
("checking Ceph config..");
477 my $conf = PVE
::Cluster
::cfs_read_file
('ceph.conf');
479 my $global = $conf->{global
};
481 my $global_monhost = $global->{mon_host
} // $global->{"mon host"} // $global->{"mon-host"};
482 if (!defined($global_monhost)) {
484 "No 'mon_host' entry found in ceph config.\n It's recommended to add mon_host with"
485 ." all monitor addresses (without ports) to the global section."
489 my $ipv6 = $global->{ms_bind_ipv6
} // $global->{"ms bind ipv6"} // $global->{"ms-bind-ipv6"};
491 my $ipv4 = $global->{ms_bind_ipv4
} // $global->{"ms bind ipv4"} // $global->{"ms-bind-ipv4"};
492 if ($ipv6 eq 'true' && (!defined($ipv4) || $ipv4 ne 'false')) {
494 "'ms_bind_ipv6' is enabled but 'ms_bind_ipv4' is not disabled.\n Make sure to"
495 ." disable 'ms_bind_ipv4' for ipv6 only clusters, or add an ipv4 network to public/cluster network."
500 if (defined($global->{keyring
})) {
502 "[global] config section contains 'keyring' option, which will prevent services from"
503 ." starting with Nautilus.\n Move 'keyring' option to [client] section instead."
508 log_warn
("Empty ceph config found");
511 my $local_ceph_ver = PVE
::Ceph
::Tools
::get_local_version
(1);
512 if (defined($local_ceph_ver)) {
513 if ($local_ceph_ver <= 14) {
514 log_fail
("local Ceph version too low, at least Octopus required..");
517 log_fail
("unable to determine local Ceph version.");
521 sub check_backup_retention_settings
{
522 log_info
("Checking backup retention settings..");
526 my $node_has_retention;
528 my $maxfiles_msg = "parameter 'maxfiles' is deprecated with PVE 7.x and will be removed in a " .
529 "future version, use 'prune-backups' instead.";
532 my $confdesc = PVE
::VZDump
::Common
::get_confdesc
();
534 my $fn = "/etc/vzdump.conf";
535 my $raw = PVE
::Tools
::file_get_contents
($fn);
537 my $conf_schema = { type
=> 'object', properties
=> $confdesc, };
538 my $param = PVE
::JSONSchema
::parse_config
($conf_schema, $fn, $raw);
540 if (defined($param->{maxfiles
})) {
542 log_warn
("$fn - $maxfiles_msg");
545 $node_has_retention = defined($param->{maxfiles
}) || defined($param->{'prune-backups'});
549 log_warn
("unable to parse node's VZDump configuration - $err");
552 my $storage_cfg = PVE
::Storage
::config
();
554 for my $storeid (keys $storage_cfg->{ids
}->%*) {
555 my $scfg = $storage_cfg->{ids
}->{$storeid};
557 if (defined($scfg->{maxfiles
})) {
559 log_warn
("storage '$storeid' - $maxfiles_msg");
562 next if !$scfg->{content
}->{backup
};
563 next if defined($scfg->{maxfiles
}) || defined($scfg->{'prune-backups'});
564 next if $node_has_retention;
567 "storage '$storeid' - no backup retention settings defined - by default, since PVE 7.0"
568 ." it will no longer keep only the last backup, but all backups"
573 my $vzdump_cron = PVE
::Cluster
::cfs_read_file
('vzdump.cron');
575 # only warn once, there might be many jobs...
576 if (scalar(grep { defined($_->{maxfiles
}) } $vzdump_cron->{jobs
}->@*)) {
578 log_warn
("/etc/pve/vzdump.cron - $maxfiles_msg");
583 log_warn
("unable to parse node's VZDump configuration - $err");
586 log_pass
("no problems found.") if $pass;
589 sub check_cifs_credential_location
{
590 log_info
("checking CIFS credential location..");
592 my $regex = qr/^(.*)\.cred$/;
596 PVE
::Tools
::dir_glob_foreach
('/etc/pve/priv/', $regex, sub {
599 my ($basename) = $filename =~ $regex;
602 "CIFS credentials '/etc/pve/priv/$filename' will be moved to"
603 ." '/etc/pve/priv/storage/$basename.pw' during the update"
609 log_pass
("no CIFS credentials at outdated location found.") if !$found;
612 sub check_custom_pool_roles
{
613 log_info
("Checking custom roles for pool permissions..");
615 if (! -f
"/etc/pve/user.cfg") {
616 log_skip
("user.cfg does not exist");
620 my $raw = eval { PVE
::Tools
::file_get_contents
('/etc/pve/user.cfg'); };
622 log_fail
("Failed to read '/etc/pve/user.cfg' - $@");
627 while ($raw =~ /^\s*(.+?)\s*$/gm) {
631 foreach my $d (split (/:/, $line)) {
637 my $et = shift @data;
638 next if $et ne 'role';
640 my ($role, $privlist) = @data;
641 if (!PVE
::AccessControl
::verify_rolename
($role, 1)) {
642 warn "user config - ignore role '$role' - invalid characters in role name\n";
646 $roles->{$role} = {} if !$roles->{$role};
647 foreach my $priv (split_list
($privlist)) {
648 $roles->{$role}->{$priv} = 1;
652 foreach my $role (sort keys %{$roles}) {
653 next if PVE
::AccessControl
::role_is_special
($role);
655 # TODO: any role updates?
659 my sub check_max_length
{
660 my ($raw, $max_length, $warning) = @_;
661 log_warn
($warning) if defined($raw) && length($raw) > $max_length;
664 sub check_node_and_guest_configurations
{
665 log_info
("Checking node and guest description/note legnth..");
667 my @affected_nodes = grep {
668 my $desc = PVE
::NodeConfig
::load_config
($_)->{desc
};
669 defined($desc) && length($desc) > 64 * 1024
670 } PVE
::Cluster
::get_nodelist
();
672 if (scalar(@affected_nodes) > 0) {
673 log_warn
("Node config description of the following nodes too long for new limit of 64 KiB:\n "
674 . join(', ', @affected_nodes));
676 log_pass
("All node config descriptions fit in the new limit of 64 KiB");
679 my $affected_guests_long_desc = [];
680 my $affected_cts_cgroup_keys = [];
682 my $cts = PVE
::LXC
::config_list
();
683 for my $vmid (sort { $a <=> $b } keys %$cts) {
684 my $conf = PVE
::LXC
::Config-
>load_config($vmid);
686 my $desc = $conf->{description
};
687 push @$affected_guests_long_desc, "CT $vmid" if defined($desc) && length($desc) > 8 * 1024;
689 my $lxc_raw_conf = $conf->{lxc
};
690 push @$affected_cts_cgroup_keys, "CT $vmid" if (grep (@$_[0] =~ /^lxc\.cgroup\./, @$lxc_raw_conf));
692 my $vms = PVE
::QemuServer
::config_list
();
693 for my $vmid (sort { $a <=> $b } keys %$vms) {
694 my $desc = PVE
::QemuConfig-
>load_config($vmid)->{description
};
695 push @$affected_guests_long_desc, "VM $vmid" if defined($desc) && length($desc) > 8 * 1024;
697 if (scalar($affected_guests_long_desc->@*) > 0) {
698 log_warn
("Guest config description of the following virtual-guests too long for new limit of 64 KiB:\n"
699 ." " . join(", ", $affected_guests_long_desc->@*));
701 log_pass
("All guest config descriptions fit in the new limit of 8 KiB");
704 log_info
("Checking container configs for deprecated lxc.cgroup entries");
706 if (scalar($affected_cts_cgroup_keys->@*) > 0) {
707 if ($forced_legacy_cgroup) {
708 log_pass
("Found legacy 'lxc.cgroup' keys, but system explicitly configured for legacy hybrid cgroup hierarchy.");
710 log_warn
("The following CTs have 'lxc.cgroup' keys configured, which will be ignored in the new default unified cgroupv2:\n"
711 ." " . join(", ", $affected_cts_cgroup_keys->@*) ."\n"
712 ." Often it can be enough to change to the new 'lxc.cgroup2' prefix after the upgrade to Proxmox VE 7.x");
715 log_pass
("No legacy 'lxc.cgroup' keys found.");
719 sub check_storage_content
{
720 log_info
("Checking storage content type configuration..");
725 my $storage_cfg = PVE
::Storage
::config
();
727 for my $storeid (sort keys $storage_cfg->{ids
}->%*) {
728 my $scfg = $storage_cfg->{ids
}->{$storeid};
730 next if $scfg->{shared
};
731 next if !PVE
::Storage
::storage_check_enabled
($storage_cfg, $storeid, undef, 1);
733 my $valid_content = PVE
::Storage
::Plugin
::valid_content_types
($scfg->{type
});
735 if (scalar(keys $scfg->{content
}->%*) == 0 && !$valid_content->{none
}) {
737 log_fail
("storage '$storeid' does not support configured content type 'none'");
738 delete $scfg->{content
}->{none
}; # scan for guest images below
741 next if $scfg->{content
}->{images
};
742 next if $scfg->{content
}->{rootdir
};
744 # Skip 'iscsi(direct)' (and foreign plugins with potentially similiar behavior) with 'none',
745 # because that means "use LUNs directly" and vdisk_list() in PVE 6.x still lists those.
746 # It's enough to *not* skip 'dir', because it is the only other storage that supports 'none'
747 # and 'images' or 'rootdir', hence being potentially misconfigured.
748 next if $scfg->{type
} ne 'dir' && $scfg->{content
}->{none
};
750 eval { PVE
::Storage
::activate_storage
($storage_cfg, $storeid) };
752 log_warn
("activating '$storeid' failed - $err");
756 my $res = eval { PVE
::Storage
::vdisk_list
($storage_cfg, $storeid); };
758 log_warn
("listing images on '$storeid' failed - $err");
761 my @volids = map { $_->{volid
} } $res->{$storeid}->@*;
763 my $number = scalar(@volids);
766 "storage '$storeid' - neither content type 'images' nor 'rootdir' configured, but"
767 ."found $number guest volume(s)"
772 my $check_volid = sub {
773 my ($volid, $vmid, $vmtype, $reference) = @_;
775 my $guesttext = $vmtype eq 'qemu' ?
'VM' : 'CT';
776 my $prefix = "$guesttext $vmid - volume '$volid' ($reference)";
778 my ($storeid) = PVE
::Storage
::parse_volume_id
($volid, 1);
779 return if !defined($storeid);
781 my $scfg = $storage_cfg->{ids
}->{$storeid};
784 log_warn
("$prefix - storage does not exist!");
788 # cannot use parse_volname for containers, as it can return 'images'
789 # but containers cannot have ISO images attached, so assume 'rootdir'
790 my $vtype = 'rootdir';
791 if ($vmtype eq 'qemu') {
792 ($vtype) = eval { PVE
::Storage
::parse_volname
($storage_cfg, $volid); };
796 if (!$scfg->{content
}->{$vtype}) {
799 log_warn
("$prefix - storage does not have content type '$vtype' configured.");
803 my $cts = PVE
::LXC
::config_list
();
804 for my $vmid (sort { $a <=> $b } keys %$cts) {
805 my $conf = PVE
::LXC
::Config-
>load_config($vmid);
810 my ($ms, $mountpoint, $reference) = @_;
812 my $volid = $mountpoint->{volume
};
813 return if !$volid || $mountpoint->{type
} ne 'volume';
815 return if $volhash->{$volid}; # volume might be referenced multiple times
817 $volhash->{$volid} = 1;
819 $check_volid->($volid, $vmid, 'lxc', $reference);
822 my $opts = { include_unused
=> 1 };
823 PVE
::LXC
::Config-
>foreach_volume_full($conf, $opts, $check, 'in config');
824 for my $snapname (keys $conf->{snapshots
}->%*) {
825 my $snap = $conf->{snapshots
}->{$snapname};
826 PVE
::LXC
::Config-
>foreach_volume_full($snap, $opts, $check, "in snapshot '$snapname'");
830 my $vms = PVE
::QemuServer
::config_list
();
831 for my $vmid (sort { $a <=> $b } keys %$vms) {
832 my $conf = PVE
::QemuConfig-
>load_config($vmid);
837 my ($key, $drive, $reference) = @_;
839 my $volid = $drive->{file
};
840 return if $volid =~ m
|^/|;
841 return if $volhash->{$volid}; # volume might be referenced multiple times
843 $volhash->{$volid} = 1;
844 $check_volid->($volid, $vmid, 'qemu', $reference);
848 extra_keys
=> ['vmstate'],
851 # startup from a suspended state works even without 'images' content type on the
852 # state storage, so do not check 'vmstate' for $conf
853 PVE
::QemuConfig-
>foreach_volume_full($conf, { include_unused
=> 1 }, $check, 'in config');
854 for my $snapname (keys $conf->{snapshots
}->%*) {
855 my $snap = $conf->{snapshots
}->{$snapname};
856 PVE
::QemuConfig-
>foreach_volume_full($snap, $opts, $check, "in snapshot '$snapname'");
861 log_warn
("Proxmox VE enforces stricter content type checks since 7.0. The guests above " .
862 "might not work until the storage configuration is fixed.");
866 log_pass
("no problems found");
870 sub check_containers_cgroup_compat
{
871 if ($forced_legacy_cgroup) {
872 log_warn
("System explicitly configured for legacy hybrid cgroup hierarchy.\n"
873 ." NOTE: support for the hybrid cgroup hierachy will be removed in future Proxmox VE 9 (~ 2025)."
877 my $supports_cgroupv2 = sub {
878 my ($conf, $rootdir, $ctid) = @_;
880 my $get_systemd_version = sub {
883 my $sd_lib_dir = -d
"/lib/systemd" ?
"/lib/systemd" : "/usr/lib/systemd";
884 my $libsd = PVE
::Tools
::dir_glob_regex
($sd_lib_dir, "libsystemd-shared-.+\.so");
885 if (defined($libsd) && $libsd =~ /libsystemd-shared-(\d+)\.so/) {
892 my $unified_cgroupv2_support = sub {
895 # https://www.freedesktop.org/software/systemd/man/systemd.html
896 # systemd is installed as symlink to /sbin/init
897 my $systemd = CORE
::readlink('/sbin/init');
899 # assume non-systemd init will run with unified cgroupv2
900 if (!defined($systemd) || $systemd !~ m
@/systemd$@) {
904 # systemd version 232 (e.g. debian stretch) supports the unified hierarchy
905 my $sdver = $get_systemd_version->();
906 if (!defined($sdver) || $sdver < 232) {
913 my $ostype = $conf->{ostype
};
914 if (!defined($ostype)) {
915 log_warn
("Found CT ($ctid) without 'ostype' set!");
916 } elsif ($ostype eq 'devuan' || $ostype eq 'alpine') {
917 return 1; # no systemd, no cgroup problems
920 my $lxc_setup = PVE
::LXC
::Setup-
>new($conf, $rootdir);
921 return $lxc_setup->protected_call($unified_cgroupv2_support);
924 my $log_problem = sub {
926 my $extra = $forced_legacy_cgroup ?
'' : " or set systemd.unified_cgroup_hierarchy=0 in the Proxmox VE hosts' kernel cmdline";
928 "Found at least one CT ($ctid) which does not support running in a unified cgroup v2 layout\n"
929 ." Consider upgrading the Containers distro${extra}! Skipping further CT compat checks."
933 my $cts = eval { PVE
::API2
::LXC-
>vmlist({ node
=> $nodename }) };
935 log_warn
("Failed to retrieve information about this node's CTs - $@");
939 if (!defined($cts) || !scalar(@$cts)) {
940 log_skip
("No containers on node detected.");
944 my @running_cts = sort { $a <=> $b } grep { $_->{status
} eq 'running' } @$cts;
945 my @offline_cts = sort { $a <=> $b } grep { $_->{status
} ne 'running' } @$cts;
947 for my $ct (@running_cts) {
948 my $ctid = $ct->{vmid
};
949 my $pid = eval { PVE
::LXC
::find_lxc_pid
($ctid) };
951 log_warn
("Failed to get PID for running CT $ctid - $err");
954 my $rootdir = "/proc/$pid/root";
955 my $conf = PVE
::LXC
::Config-
>load_config($ctid);
957 my $ret = eval { $supports_cgroupv2->($conf, $rootdir, $ctid) };
959 log_warn
("Failed to get cgroup support status for CT $ctid - $err");
963 $log_problem->($ctid);
968 my $storage_cfg = PVE
::Storage
::config
();
969 for my $ct (@offline_cts) {
970 my $ctid = $ct->{vmid
};
971 my ($conf, $rootdir, $ret);
973 $conf = PVE
::LXC
::Config-
>load_config($ctid);
974 $rootdir = PVE
::LXC
::mount_all
($ctid, $storage_cfg, $conf);
975 $ret = $supports_cgroupv2->($conf, $rootdir, $ctid);
978 log_warn
("Failed to load config and mount CT $ctid - $err");
979 eval { PVE
::LXC
::umount_all
($ctid, $storage_cfg, $conf) };
983 $log_problem->($ctid);
984 eval { PVE
::LXC
::umount_all
($ctid, $storage_cfg, $conf) };
988 eval { PVE
::LXC
::umount_all
($ctid, $storage_cfg, $conf) };
992 sub check_security_repo
{
993 log_info
("Checking if the suite for the Debian security repository is correct..");
997 my $dir = '/etc/apt/sources.list.d';
1000 my $check_file = sub {
1003 $file = "${dir}/${file}" if $in_dir;
1005 my $raw = eval { PVE
::Tools
::file_get_contents
($file) };
1006 return if !defined($raw);
1007 my @lines = split(/\n/, $raw);
1010 for my $line (@lines) {
1013 next if length($line) == 0; # split would result in undef then...
1015 ($line) = split(/#/, $line);
1017 next if $line !~ m/^deb[[:space:]]/; # is case sensitive
1022 # https://deb.debian.org/debian-security
1023 # http://security.debian.org/debian-security
1024 # http://security.debian.org/
1025 if ($line =~ m
|https?
://deb\
.debian\
.org
/debian-security/?\s
+(\S
*)|i
) {
1027 } elsif ($line =~ m
|https?
://security\
.debian\
.org
(?
:.*?
)\s
+(\S
*)|i
) {
1035 my $where = "in ${file}:${number}";
1037 if ($suite eq 'buster/updates') {
1038 log_info
("Make sure to change the suite of the Debian security repository " .
1039 "from 'buster/updates' to 'bullseye-security' - $where");
1040 } elsif ($suite eq 'bullseye-security') {
1041 log_pass
("already using 'bullseye-security'");
1043 log_fail
("The new suite of the Debian security repository should be " .
1044 "'bullseye-security' - $where");
1049 $check_file->("/etc/apt/sources.list");
1053 PVE
::Tools
::dir_glob_foreach
($dir, '^.*\.list$', $check_file);
1056 # only warn, it might be defined in a .sources file or in a way not catched above
1057 log_warn
("No Debian security repository detected in /etc/apt/sources.list and " .
1058 "/etc/apt/sources.list.d/*.list");
1063 print_header
("MISCELLANEOUS CHECKS");
1064 my $ssh_config = eval { PVE
::Tools
::file_get_contents
('/root/.ssh/config') };
1065 if (defined($ssh_config)) {
1066 log_fail
("Unsupported SSH Cipher configured for root in /root/.ssh/config: $1")
1067 if $ssh_config =~ /^Ciphers .*(blowfish|arcfour|3des).*$/m;
1069 log_skip
("No SSH config file found.");
1072 log_info
("Checking common daemon services..");
1073 $log_systemd_unit_state->('pveproxy.service');
1074 $log_systemd_unit_state->('pvedaemon.service');
1075 $log_systemd_unit_state->('pvescheduler.service');
1076 $log_systemd_unit_state->('pvestatd.service');
1078 my $root_free = PVE
::Tools
::df
('/', 10);
1079 log_warn
("Less than 5 GB free space on root file system.")
1080 if defined($root_free) && $root_free->{avail
} < 5 * 1000*1000*1000;
1082 log_info
("Checking for running guests..");
1083 my $running_guests = 0;
1085 my $vms = eval { PVE
::API2
::Qemu-
>vmlist({ node
=> $nodename }) };
1086 log_warn
("Failed to retrieve information about this node's VMs - $@") if $@;
1087 $running_guests += grep { $_->{status
} eq 'running' } @$vms if defined($vms);
1089 my $cts = eval { PVE
::API2
::LXC-
>vmlist({ node
=> $nodename }) };
1090 log_warn
("Failed to retrieve information about this node's CTs - $@") if $@;
1091 $running_guests += grep { $_->{status
} eq 'running' } @$cts if defined($cts);
1093 if ($running_guests > 0) {
1094 log_warn
("$running_guests running guest(s) detected - consider migrating or stopping them.")
1096 log_pass
("no running guest detected.")
1099 log_info
("Checking if the local node's hostname '$nodename' is resolvable..");
1100 my $local_ip = eval { PVE
::Network
::get_ip_from_hostname
($nodename) };
1102 log_warn
("Failed to resolve hostname '$nodename' to IP - $@");
1104 log_info
("Checking if resolved IP is configured on local node..");
1105 my $cidr = Net
::IP
::ip_is_ipv6
($local_ip) ?
"$local_ip/128" : "$local_ip/32";
1106 my $configured_ips = PVE
::Network
::get_local_ip_from_cidr
($cidr);
1107 my $ip_count = scalar(@$configured_ips);
1109 if ($ip_count <= 0) {
1110 log_fail
("Resolved node IP '$local_ip' not configured or active for '$nodename'");
1111 } elsif ($ip_count > 1) {
1112 log_warn
("Resolved node IP '$local_ip' active on multiple ($ip_count) interfaces!");
1114 log_pass
("Resolved node IP '$local_ip' configured and active on single interface.");
1118 log_info
("Check node certificate's RSA key size");
1119 my $certs = PVE
::API2
::Certificates-
>info({ node
=> $nodename });
1121 'rsaEncryption' => {
1125 'id-ecPublicKey' => {
1131 my $certs_check_failed = 0;
1132 foreach my $cert (@$certs) {
1133 my ($type, $size, $fn) = $cert->@{qw(public-key-type public-key-bits filename)};
1135 if (!defined($type) || !defined($size)) {
1136 log_warn
("'$fn': cannot check certificate, failed to get it's type or size!");
1139 my $check = $certs_check->{$type};
1140 if (!defined($check)) {
1141 log_warn
("'$fn': certificate's public key type '$type' unknown!");
1145 if ($size < $check->{minsize
}) {
1146 log_fail
("'$fn', certificate's $check->{name} public key size is less than 2048 bit");
1147 $certs_check_failed = 1;
1149 log_pass
("Certificate '$fn' passed Debian Busters (and newer) security level for TLS connections ($size >= 2048)");
1153 check_backup_retention_settings
();
1154 check_cifs_credential_location
();
1155 check_custom_pool_roles
();
1156 check_node_and_guest_configurations
();
1157 check_security_repo
();
1160 __PACKAGE__-
>register_method ({
1161 name
=> 'checklist',
1162 path
=> 'checklist',
1164 description
=> 'Check (pre-/post-)upgrade conditions.',
1166 additionalProperties
=> 0,
1169 description
=> 'perform additional, expensive checks.',
1176 returns
=> { type
=> 'null' },
1180 my $kernel_cli = PVE
::Tools
::file_get_contents
('/proc/cmdline');
1181 if ($kernel_cli =~ /systemd.unified_cgroup_hierarchy=0/){
1182 $forced_legacy_cgroup = 1;
1185 check_pve_packages
();
1186 check_cluster_corosync
();
1188 check_storage_health
();
1191 if ($param->{full
}) {
1192 check_containers_cgroup_compat
();
1194 log_skip
("NOTE: Expensive checks, like CT cgroupv2 compat, not performed without '--full' parameter");
1197 print_header
("SUMMARY");
1200 $total += $_ for values %$counters;
1202 print "TOTAL: $total\n";
1203 print colored
("PASSED: $counters->{pass}\n", 'green');
1204 print "SKIPPED: $counters->{skip}\n";
1205 print colored
("WARNINGS: $counters->{warn}\n", 'yellow');
1206 print colored
("FAILURES: $counters->{fail}\n", 'red');
1208 if ($counters->{warn} > 0 || $counters->{fail
} > 0) {
1209 my $color = $counters->{fail
} > 0 ?
'red' : 'yellow';
1210 print colored
("\nATTENTION: Please check the output for detailed information!\n", $color);
1211 print colored
("Try to solve the problems one at a time and then run this checklist tool again.\n", $color) if $counters->{fail
} > 0;
1217 our $cmddef = [ __PACKAGE__
, 'checklist', [], {}];