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 $get_systemd_unit_state = sub {
95 my ($unit, $surpress_stderr) = @_;
98 my $filter_output = sub {
103 my %extra = (outfunc
=> $filter_output, noerr
=> 1);
104 $extra{errfunc
} = sub { } if $surpress_stderr;
107 run_command
(['systemctl', 'is-enabled', "$unit"], %extra);
108 return if !defined($state);
109 run_command
(['systemctl', 'is-active', "$unit"], %extra);
112 return $state // 'unknown';
114 my $log_systemd_unit_state = sub {
115 my ($unit, $no_fail_on_inactive) = @_;
117 my $log_method = \
&log_warn
;
119 my $state = $get_systemd_unit_state->($unit);
120 if ($state eq 'active') {
121 $log_method = \
&log_pass
;
122 } elsif ($state eq 'inactive') {
123 $log_method = $no_fail_on_inactive ? \
&log_warn
: \
&log_fail
;
124 } elsif ($state eq 'failed') {
125 $log_method = \
&log_fail
;
128 $log_method->("systemd unit '$unit' is in state '$state'");
135 $versions = eval { PVE
::API2
::APT-
>versions({ node
=> $nodename }) } if !defined($versions);
137 if (!defined($versions)) {
138 my $msg = "unable to retrieve package version information";
139 $msg .= "- $@" if $@;
144 my $pkgs = [ grep { $_->{Package
} eq $pkg } @$versions ];
145 if (!defined $pkgs || $pkgs == 0) {
146 log_fail
("unable to determine installed $pkg version.");
153 sub check_pve_packages
{
154 print_header
("CHECKING VERSION INFORMATION FOR PVE PACKAGES");
156 print "Checking for package updates..\n";
157 my $updates = eval { PVE
::API2
::APT-
>list_updates({ node
=> $nodename }); };
158 if (!defined($updates)) {
159 log_warn
("$@") if $@;
160 log_fail
("unable to retrieve list of package updates!");
161 } elsif (@$updates > 0) {
162 my $pkgs = join(', ', map { $_->{Package
} } @$updates);
163 log_warn
("updates for the following packages are available:\n $pkgs");
165 log_pass
("all packages uptodate");
168 print "\nChecking proxmox-ve package version..\n";
169 if (defined(my $proxmox_ve = $get_pkg->('proxmox-ve'))) {
170 my $min_pve_ver = "$min_pve_major.$min_pve_minor-$min_pve_pkgrel";
172 my ($maj, $min, $pkgrel) = $proxmox_ve->{OldVersion
} =~ m/^(\d+)\.(\d+)-(\d+)/;
176 if ($maj > $min_pve_major) {
177 log_pass
("already upgraded to Proxmox VE " . ($min_pve_major + 1));
179 } elsif ($maj >= $min_pve_major && $min >= $min_pve_minor && $pkgrel >= $min_pve_pkgrel) {
180 log_pass
("proxmox-ve package has version >= $min_pve_ver");
182 log_fail
("proxmox-ve package is too old, please upgrade to >= $min_pve_ver!");
185 my ($krunning, $kinstalled) = (qr/6\.(?:2|5)/, 'pve-kernel-6.2');
187 # we got a few that avoided 5.15 in cluster with mixed CPUs, so allow older too
188 ($krunning, $kinstalled) = (qr/(?:5\.(?:13|15)|6\.2)/, 'pve-kernel-5.15');
191 print "\nChecking running kernel version..\n";
192 my $kernel_ver = $proxmox_ve->{RunningKernel
};
193 if (!defined($kernel_ver)) {
194 log_fail
("unable to determine running kernel version.");
195 } elsif ($kernel_ver =~ /^$krunning/) {
197 log_pass
("running new kernel '$kernel_ver' after upgrade.");
199 log_pass
("running kernel '$kernel_ver' is considered suitable for upgrade.");
201 } elsif ($get_pkg->($kinstalled)) {
202 # with 6.2 kernel being available in both we might want to fine-tune the check?
203 log_warn
("a suitable kernel ($kinstalled) is intalled, but an unsuitable ($kernel_ver) is booted, missing reboot?!");
205 log_warn
("unexpected running and installed kernel '$kernel_ver'.");
208 log_fail
("proxmox-ve package not found!");
213 sub check_storage_health
{
214 print_header
("CHECKING CONFIGURED STORAGES");
215 my $cfg = PVE
::Storage
::config
();
219 my $info = PVE
::Storage
::storage_info
($cfg);
221 foreach my $storeid (sort keys %$info) {
222 my $d = $info->{$storeid};
225 log_pass
("storage '$storeid' enabled and active.");
227 log_warn
("storage '$storeid' enabled but not active!");
230 log_skip
("storage '$storeid' disabled.");
234 check_storage_content
();
237 sub check_cluster_corosync
{
238 print_header
("CHECKING CLUSTER HEALTH/SETTINGS");
240 if (!PVE
::Corosync
::check_conf_exists
(1)) {
241 log_skip
("standalone node.");
245 $log_systemd_unit_state->('pve-cluster.service');
246 $log_systemd_unit_state->('corosync.service');
248 if (PVE
::Cluster
::check_cfs_quorum
(1)) {
249 log_pass
("Cluster Filesystem is quorate.");
251 log_fail
("Cluster Filesystem readonly, lost quorum?!");
254 my $conf = PVE
::Cluster
::cfs_read_file
('corosync.conf');
255 my $conf_nodelist = PVE
::Corosync
::nodelist
($conf);
258 print "\nAnalzying quorum settings and state..\n";
259 if (!defined($conf_nodelist)) {
260 log_fail
("unable to retrieve nodelist from corosync.conf");
262 if (grep { $conf_nodelist->{$_}->{quorum_votes
} != 1 } keys %$conf_nodelist) {
263 log_warn
("non-default quorum_votes distribution detected!");
265 map { $node_votes += $conf_nodelist->{$_}->{quorum_votes
} // 0 } keys %$conf_nodelist;
268 my ($expected_votes, $total_votes);
269 my $filter_output = sub {
271 ($expected_votes) = $line =~ /^Expected votes:\s*(\d+)\s*$/
272 if !defined($expected_votes);
273 ($total_votes) = $line =~ /^Total votes:\s*(\d+)\s*$/
274 if !defined($total_votes);
277 run_command
(['corosync-quorumtool', '-s'], outfunc
=> $filter_output, noerr
=> 1);
280 if (!defined($expected_votes)) {
281 log_fail
("unable to get expected number of votes, assuming 0.");
284 if (!defined($total_votes)) {
285 log_fail
("unable to get expected number of votes, assuming 0.");
289 my $cfs_nodelist = PVE
::Cluster
::get_clinfo
()->{nodelist
};
290 my $offline_nodes = grep { $cfs_nodelist->{$_}->{online
} != 1 } keys %$cfs_nodelist;
291 if ($offline_nodes > 0) {
292 log_fail
("$offline_nodes nodes are offline!");
295 my $qdevice_votes = 0;
296 if (my $qdevice_setup = $conf->{main
}->{quorum
}->{device
}) {
297 $qdevice_votes = $qdevice_setup->{votes
} // 1;
300 log_info
("configured votes - nodes: $node_votes");
301 log_info
("configured votes - qdevice: $qdevice_votes");
302 log_info
("current expected votes: $expected_votes");
303 log_info
("current total votes: $total_votes");
305 log_warn
("expected votes set to non-standard value '$expected_votes'.")
306 if $expected_votes != $node_votes + $qdevice_votes;
307 log_warn
("total votes < expected votes: $total_votes/$expected_votes!")
308 if $total_votes < $expected_votes;
310 my $conf_nodelist_count = scalar(keys %$conf_nodelist);
311 my $cfs_nodelist_count = scalar(keys %$cfs_nodelist);
312 log_warn
("cluster consists of less than three quorum-providing nodes!")
313 if $conf_nodelist_count < 3 && $conf_nodelist_count + $qdevice_votes < 3;
315 log_fail
("corosync.conf ($conf_nodelist_count) and pmxcfs ($cfs_nodelist_count) don't agree about size of nodelist.")
316 if $conf_nodelist_count != $cfs_nodelist_count;
318 print "\nChecking nodelist entries..\n";
319 my $nodelist_pass = 1;
320 for my $cs_node (sort keys %$conf_nodelist) {
321 my $entry = $conf_nodelist->{$cs_node};
322 if (!defined($entry->{name
})) {
324 log_fail
("$cs_node: no name entry in corosync.conf.");
326 if (!defined($entry->{nodeid
})) {
328 log_fail
("$cs_node: no nodeid configured in corosync.conf.");
331 for my $link (0..7) {
332 $gotLinks++ if defined($entry->{"ring${link}_addr"});
334 if ($gotLinks <= 0) {
336 log_fail
("$cs_node: no ringX_addr (0 <= X <= 7) link defined in corosync.conf.");
339 my $verify_ring_ip = sub {
341 if (defined(my $ring = $entry->{$key})) {
342 my ($resolved_ip, undef) = PVE
::Corosync
::resolve_hostname_like_corosync
($ring, $conf);
343 if (defined($resolved_ip)) {
344 if ($resolved_ip ne $ring) {
347 "$cs_node: $key '$ring' resolves to '$resolved_ip'.\n"
348 ." Consider replacing it with the currently resolved IP address."
354 "$cs_node: unable to resolve $key '$ring' to an IP address according to Corosync's"
355 ." resolve strategy - cluster will potentially fail with Corosync 3.x/kronosnet!"
360 for my $link (0..7) {
361 $verify_ring_ip->("ring${link}_addr");
364 log_pass
("nodelist settings OK") if $nodelist_pass;
366 print "\nChecking totem settings..\n";
367 my $totem = $conf->{main
}->{totem
};
370 my $transport = $totem->{transport
};
371 if (defined($transport)) {
372 if ($transport ne 'knet') {
374 log_fail
("Corosync transport explicitly set to '$transport' instead of implicit default!");
378 # TODO: are those values still up-to-date?
379 if ((!defined($totem->{secauth
}) || $totem->{secauth
} ne 'on') && (!defined($totem->{crypto_cipher
}) || $totem->{crypto_cipher
} eq 'none')) {
381 log_fail
("Corosync authentication/encryption is not explicitly enabled (secauth / crypto_cipher / crypto_hash)!");
382 } elsif (defined($totem->{crypto_cipher
}) && $totem->{crypto_cipher
} eq '3des') {
384 log_fail
("Corosync encryption cipher set to '3des', no longer supported in Corosync 3.x!"); # FIXME: can be removed?
387 log_pass
("totem settings OK") if $totem_pass;
389 log_info
("run 'pvecm status' to get detailed cluster status..");
391 if (defined(my $corosync = $get_pkg->('corosync'))) {
392 if ($corosync->{OldVersion
} =~ m/^2\./) {
393 log_fail
("\ncorosync 2.x installed, cluster-wide upgrade to 3.x needed!");
394 } elsif ($corosync->{OldVersion
} !~ m/^3\./) {
395 log_fail
("\nunexpected corosync version installed: $corosync->{OldVersion}!");
401 print_header
("CHECKING HYPER-CONVERGED CEPH STATUS");
403 if (PVE
::Ceph
::Tools
::check_ceph_inited
(1)) {
404 log_info
("hyper-converged ceph setup detected!");
406 log_skip
("no hyper-converged ceph setup detected!");
410 log_info
("getting Ceph status/health information..");
411 my $ceph_status = eval { PVE
::API2
::Ceph-
>status({ node
=> $nodename }); };
412 my $noout = eval { PVE
::API2
::Cluster
::Ceph-
>get_flag({ flag
=> "noout" }); };
414 log_fail
("failed to get 'noout' flag status - $@");
417 my $noout_wanted = 1;
419 if (!$ceph_status || !$ceph_status->{health
}) {
420 log_fail
("unable to determine Ceph status!");
422 my $ceph_health = $ceph_status->{health
}->{status
};
424 log_fail
("unable to determine Ceph health!");
425 } elsif ($ceph_health eq 'HEALTH_OK') {
426 log_pass
("Ceph health reported as 'HEALTH_OK'.");
427 } elsif ($ceph_health eq 'HEALTH_WARN' && $noout && (keys %{$ceph_status->{health
}->{checks
}} == 1)) {
428 log_pass
("Ceph health reported as 'HEALTH_WARN' with a single failing check and 'noout' flag set.");
430 log_warn
("Ceph health reported as '$ceph_health'.\n Use the PVE ".
431 "dashboard or 'ceph -s' to determine the specific issues and try to resolve them.");
435 # TODO: check OSD min-required version, if to low it breaks stuff!
437 log_info
("getting Ceph daemon versions..");
438 my $ceph_versions = eval { PVE
::Ceph
::Tools
::get_cluster_versions
(undef, 1); };
439 if (!$ceph_versions) {
440 log_fail
("unable to determine Ceph daemon versions!");
443 { 'key' => 'mon', 'name' => 'monitor' },
444 { 'key' => 'mgr', 'name' => 'manager' },
445 { 'key' => 'mds', 'name' => 'MDS' },
446 { 'key' => 'osd', 'name' => 'OSD' },
449 foreach my $service (@$services) {
450 my $name = $service->{name
};
451 if (my $service_versions = $ceph_versions->{$service->{key
}}) {
452 if (keys %$service_versions == 0) {
453 log_skip
("no running instances detected for daemon type $name.");
454 } elsif (keys %$service_versions == 1) {
455 log_pass
("single running version detected for daemon type $name.");
457 log_warn
("multiple running versions detected for daemon type $name!");
460 log_skip
("unable to determine versions of running Ceph $name instances.");
464 my $overall_versions = $ceph_versions->{overall
};
465 if (!$overall_versions) {
466 log_warn
("unable to determine overall Ceph daemon versions!");
467 } elsif (keys %$overall_versions == 1) {
468 log_pass
("single running overall version detected for all Ceph daemon types.");
469 $noout_wanted = 0; # off post-upgrade, on pre-upgrade
471 log_warn
("overall version mismatch detected, check 'ceph versions' output for details!");
477 log_pass
("'noout' flag set to prevent rebalancing during cluster-wide upgrades.");
479 log_warn
("'noout' flag set, Ceph cluster upgrade seems finished.");
481 } elsif ($noout_wanted) {
482 log_warn
("'noout' flag not set - recommended to prevent rebalancing during upgrades.");
485 log_info
("checking Ceph config..");
486 my $conf = PVE
::Cluster
::cfs_read_file
('ceph.conf');
488 my $global = $conf->{global
};
490 my $global_monhost = $global->{mon_host
} // $global->{"mon host"} // $global->{"mon-host"};
491 if (!defined($global_monhost)) {
493 "No 'mon_host' entry found in ceph config.\n It's recommended to add mon_host with"
494 ." all monitor addresses (without ports) to the global section."
498 my $ipv6 = $global->{ms_bind_ipv6
} // $global->{"ms bind ipv6"} // $global->{"ms-bind-ipv6"};
500 my $ipv4 = $global->{ms_bind_ipv4
} // $global->{"ms bind ipv4"} // $global->{"ms-bind-ipv4"};
501 if ($ipv6 eq 'true' && (!defined($ipv4) || $ipv4 ne 'false')) {
503 "'ms_bind_ipv6' is enabled but 'ms_bind_ipv4' is not disabled.\n Make sure to"
504 ." disable 'ms_bind_ipv4' for ipv6 only clusters, or add an ipv4 network to public/cluster network."
509 if (defined($global->{keyring
})) {
511 "[global] config section contains 'keyring' option, which will prevent services from"
512 ." starting with Nautilus.\n Move 'keyring' option to [client] section instead."
517 log_warn
("Empty ceph config found");
520 my $local_ceph_ver = PVE
::Ceph
::Tools
::get_local_version
(1);
521 if (defined($local_ceph_ver)) {
522 if ($local_ceph_ver <= 14) {
523 log_fail
("local Ceph version too low, at least Octopus required..");
526 log_fail
("unable to determine local Ceph version.");
530 sub check_backup_retention_settings
{
531 log_info
("Checking backup retention settings..");
535 my $node_has_retention;
537 my $maxfiles_msg = "parameter 'maxfiles' is deprecated with PVE 7.x and will be removed in a " .
538 "future version, use 'prune-backups' instead.";
541 my $confdesc = PVE
::VZDump
::Common
::get_confdesc
();
543 my $fn = "/etc/vzdump.conf";
544 my $raw = PVE
::Tools
::file_get_contents
($fn);
546 my $conf_schema = { type
=> 'object', properties
=> $confdesc, };
547 my $param = PVE
::JSONSchema
::parse_config
($conf_schema, $fn, $raw);
549 if (defined($param->{maxfiles
})) {
551 log_warn
("$fn - $maxfiles_msg");
554 $node_has_retention = defined($param->{maxfiles
}) || defined($param->{'prune-backups'});
558 log_warn
("unable to parse node's VZDump configuration - $err");
561 my $storage_cfg = PVE
::Storage
::config
();
563 for my $storeid (keys $storage_cfg->{ids
}->%*) {
564 my $scfg = $storage_cfg->{ids
}->{$storeid};
566 if (defined($scfg->{maxfiles
})) {
568 log_warn
("storage '$storeid' - $maxfiles_msg");
571 next if !$scfg->{content
}->{backup
};
572 next if defined($scfg->{maxfiles
}) || defined($scfg->{'prune-backups'});
573 next if $node_has_retention;
576 "storage '$storeid' - no backup retention settings defined - by default, since PVE 7.0"
577 ." it will no longer keep only the last backup, but all backups"
582 my $vzdump_cron = PVE
::Cluster
::cfs_read_file
('vzdump.cron');
584 # only warn once, there might be many jobs...
585 if (scalar(grep { defined($_->{maxfiles
}) } $vzdump_cron->{jobs
}->@*)) {
587 log_warn
("/etc/pve/vzdump.cron - $maxfiles_msg");
592 log_warn
("unable to parse node's VZDump configuration - $err");
595 log_pass
("no problems found.") if $pass;
598 sub check_cifs_credential_location
{
599 log_info
("checking CIFS credential location..");
601 my $regex = qr/^(.*)\.cred$/;
605 PVE
::Tools
::dir_glob_foreach
('/etc/pve/priv/', $regex, sub {
608 my ($basename) = $filename =~ $regex;
611 "CIFS credentials '/etc/pve/priv/$filename' will be moved to"
612 ." '/etc/pve/priv/storage/$basename.pw' during the update"
618 log_pass
("no CIFS credentials at outdated location found.") if !$found;
621 sub check_custom_pool_roles
{
622 log_info
("Checking custom roles for pool permissions..");
624 if (! -f
"/etc/pve/user.cfg") {
625 log_skip
("user.cfg does not exist");
629 my $raw = eval { PVE
::Tools
::file_get_contents
('/etc/pve/user.cfg'); };
631 log_fail
("Failed to read '/etc/pve/user.cfg' - $@");
636 while ($raw =~ /^\s*(.+?)\s*$/gm) {
640 foreach my $d (split (/:/, $line)) {
646 my $et = shift @data;
647 next if $et ne 'role';
649 my ($role, $privlist) = @data;
650 if (!PVE
::AccessControl
::verify_rolename
($role, 1)) {
651 warn "user config - ignore role '$role' - invalid characters in role name\n";
655 $roles->{$role} = {} if !$roles->{$role};
656 foreach my $priv (split_list
($privlist)) {
657 $roles->{$role}->{$priv} = 1;
661 foreach my $role (sort keys %{$roles}) {
662 next if PVE
::AccessControl
::role_is_special
($role);
664 # TODO: any role updates?
668 my sub check_max_length
{
669 my ($raw, $max_length, $warning) = @_;
670 log_warn
($warning) if defined($raw) && length($raw) > $max_length;
673 sub check_node_and_guest_configurations
{
674 log_info
("Checking node and guest description/note legnth..");
676 my @affected_nodes = grep {
677 my $desc = PVE
::NodeConfig
::load_config
($_)->{desc
};
678 defined($desc) && length($desc) > 64 * 1024
679 } PVE
::Cluster
::get_nodelist
();
681 if (scalar(@affected_nodes) > 0) {
682 log_warn
("Node config description of the following nodes too long for new limit of 64 KiB:\n "
683 . join(', ', @affected_nodes));
685 log_pass
("All node config descriptions fit in the new limit of 64 KiB");
688 my $affected_guests_long_desc = [];
689 my $affected_cts_cgroup_keys = [];
691 my $cts = PVE
::LXC
::config_list
();
692 for my $vmid (sort { $a <=> $b } keys %$cts) {
693 my $conf = PVE
::LXC
::Config-
>load_config($vmid);
695 my $desc = $conf->{description
};
696 push @$affected_guests_long_desc, "CT $vmid" if defined($desc) && length($desc) > 8 * 1024;
698 my $lxc_raw_conf = $conf->{lxc
};
699 push @$affected_cts_cgroup_keys, "CT $vmid" if (grep (@$_[0] =~ /^lxc\.cgroup\./, @$lxc_raw_conf));
701 my $vms = PVE
::QemuServer
::config_list
();
702 for my $vmid (sort { $a <=> $b } keys %$vms) {
703 my $desc = PVE
::QemuConfig-
>load_config($vmid)->{description
};
704 push @$affected_guests_long_desc, "VM $vmid" if defined($desc) && length($desc) > 8 * 1024;
706 if (scalar($affected_guests_long_desc->@*) > 0) {
707 log_warn
("Guest config description of the following virtual-guests too long for new limit of 64 KiB:\n"
708 ." " . join(", ", $affected_guests_long_desc->@*));
710 log_pass
("All guest config descriptions fit in the new limit of 8 KiB");
713 log_info
("Checking container configs for deprecated lxc.cgroup entries");
715 if (scalar($affected_cts_cgroup_keys->@*) > 0) {
716 if ($forced_legacy_cgroup) {
717 log_pass
("Found legacy 'lxc.cgroup' keys, but system explicitly configured for legacy hybrid cgroup hierarchy.");
719 log_warn
("The following CTs have 'lxc.cgroup' keys configured, which will be ignored in the new default unified cgroupv2:\n"
720 ." " . join(", ", $affected_cts_cgroup_keys->@*) ."\n"
721 ." Often it can be enough to change to the new 'lxc.cgroup2' prefix after the upgrade to Proxmox VE 7.x");
724 log_pass
("No legacy 'lxc.cgroup' keys found.");
728 sub check_storage_content
{
729 log_info
("Checking storage content type configuration..");
734 my $storage_cfg = PVE
::Storage
::config
();
736 for my $storeid (sort keys $storage_cfg->{ids
}->%*) {
737 my $scfg = $storage_cfg->{ids
}->{$storeid};
739 next if $scfg->{shared
};
740 next if !PVE
::Storage
::storage_check_enabled
($storage_cfg, $storeid, undef, 1);
742 my $valid_content = PVE
::Storage
::Plugin
::valid_content_types
($scfg->{type
});
744 if (scalar(keys $scfg->{content
}->%*) == 0 && !$valid_content->{none
}) {
746 log_fail
("storage '$storeid' does not support configured content type 'none'");
747 delete $scfg->{content
}->{none
}; # scan for guest images below
750 next if $scfg->{content
}->{images
};
751 next if $scfg->{content
}->{rootdir
};
753 # Skip 'iscsi(direct)' (and foreign plugins with potentially similiar behavior) with 'none',
754 # because that means "use LUNs directly" and vdisk_list() in PVE 6.x still lists those.
755 # It's enough to *not* skip 'dir', because it is the only other storage that supports 'none'
756 # and 'images' or 'rootdir', hence being potentially misconfigured.
757 next if $scfg->{type
} ne 'dir' && $scfg->{content
}->{none
};
759 eval { PVE
::Storage
::activate_storage
($storage_cfg, $storeid) };
761 log_warn
("activating '$storeid' failed - $err");
765 my $res = eval { PVE
::Storage
::vdisk_list
($storage_cfg, $storeid); };
767 log_warn
("listing images on '$storeid' failed - $err");
770 my @volids = map { $_->{volid
} } $res->{$storeid}->@*;
772 my $number = scalar(@volids);
775 "storage '$storeid' - neither content type 'images' nor 'rootdir' configured, but"
776 ."found $number guest volume(s)"
781 my $check_volid = sub {
782 my ($volid, $vmid, $vmtype, $reference) = @_;
784 my $guesttext = $vmtype eq 'qemu' ?
'VM' : 'CT';
785 my $prefix = "$guesttext $vmid - volume '$volid' ($reference)";
787 my ($storeid) = PVE
::Storage
::parse_volume_id
($volid, 1);
788 return if !defined($storeid);
790 my $scfg = $storage_cfg->{ids
}->{$storeid};
793 log_warn
("$prefix - storage does not exist!");
797 # cannot use parse_volname for containers, as it can return 'images'
798 # but containers cannot have ISO images attached, so assume 'rootdir'
799 my $vtype = 'rootdir';
800 if ($vmtype eq 'qemu') {
801 ($vtype) = eval { PVE
::Storage
::parse_volname
($storage_cfg, $volid); };
805 if (!$scfg->{content
}->{$vtype}) {
808 log_warn
("$prefix - storage does not have content type '$vtype' configured.");
812 my $cts = PVE
::LXC
::config_list
();
813 for my $vmid (sort { $a <=> $b } keys %$cts) {
814 my $conf = PVE
::LXC
::Config-
>load_config($vmid);
819 my ($ms, $mountpoint, $reference) = @_;
821 my $volid = $mountpoint->{volume
};
822 return if !$volid || $mountpoint->{type
} ne 'volume';
824 return if $volhash->{$volid}; # volume might be referenced multiple times
826 $volhash->{$volid} = 1;
828 $check_volid->($volid, $vmid, 'lxc', $reference);
831 my $opts = { include_unused
=> 1 };
832 PVE
::LXC
::Config-
>foreach_volume_full($conf, $opts, $check, 'in config');
833 for my $snapname (keys $conf->{snapshots
}->%*) {
834 my $snap = $conf->{snapshots
}->{$snapname};
835 PVE
::LXC
::Config-
>foreach_volume_full($snap, $opts, $check, "in snapshot '$snapname'");
839 my $vms = PVE
::QemuServer
::config_list
();
840 for my $vmid (sort { $a <=> $b } keys %$vms) {
841 my $conf = PVE
::QemuConfig-
>load_config($vmid);
846 my ($key, $drive, $reference) = @_;
848 my $volid = $drive->{file
};
849 return if $volid =~ m
|^/|;
850 return if $volhash->{$volid}; # volume might be referenced multiple times
852 $volhash->{$volid} = 1;
853 $check_volid->($volid, $vmid, 'qemu', $reference);
857 extra_keys
=> ['vmstate'],
860 # startup from a suspended state works even without 'images' content type on the
861 # state storage, so do not check 'vmstate' for $conf
862 PVE
::QemuConfig-
>foreach_volume_full($conf, { include_unused
=> 1 }, $check, 'in config');
863 for my $snapname (keys $conf->{snapshots
}->%*) {
864 my $snap = $conf->{snapshots
}->{$snapname};
865 PVE
::QemuConfig-
>foreach_volume_full($snap, $opts, $check, "in snapshot '$snapname'");
870 log_warn
("Proxmox VE enforces stricter content type checks since 7.0. The guests above " .
871 "might not work until the storage configuration is fixed.");
875 log_pass
("no problems found");
879 sub check_containers_cgroup_compat
{
880 if ($forced_legacy_cgroup) {
881 log_warn
("System explicitly configured for legacy hybrid cgroup hierarchy.\n"
882 ." NOTE: support for the hybrid cgroup hierachy will be removed in future Proxmox VE 9 (~ 2025)."
886 my $supports_cgroupv2 = sub {
887 my ($conf, $rootdir, $ctid) = @_;
889 my $get_systemd_version = sub {
892 my $sd_lib_dir = -d
"/lib/systemd" ?
"/lib/systemd" : "/usr/lib/systemd";
893 my $libsd = PVE
::Tools
::dir_glob_regex
($sd_lib_dir, "libsystemd-shared-.+\.so");
894 if (defined($libsd) && $libsd =~ /libsystemd-shared-(\d+)\.so/) {
901 my $unified_cgroupv2_support = sub {
904 # https://www.freedesktop.org/software/systemd/man/systemd.html
905 # systemd is installed as symlink to /sbin/init
906 my $systemd = CORE
::readlink('/sbin/init');
908 # assume non-systemd init will run with unified cgroupv2
909 if (!defined($systemd) || $systemd !~ m
@/systemd$@) {
913 # systemd version 232 (e.g. debian stretch) supports the unified hierarchy
914 my $sdver = $get_systemd_version->();
915 if (!defined($sdver) || $sdver < 232) {
922 my $ostype = $conf->{ostype
};
923 if (!defined($ostype)) {
924 log_warn
("Found CT ($ctid) without 'ostype' set!");
925 } elsif ($ostype eq 'devuan' || $ostype eq 'alpine') {
926 return 1; # no systemd, no cgroup problems
929 my $lxc_setup = PVE
::LXC
::Setup-
>new($conf, $rootdir);
930 return $lxc_setup->protected_call($unified_cgroupv2_support);
933 my $log_problem = sub {
935 my $extra = $forced_legacy_cgroup ?
'' : " or set systemd.unified_cgroup_hierarchy=0 in the Proxmox VE hosts' kernel cmdline";
937 "Found at least one CT ($ctid) which does not support running in a unified cgroup v2 layout\n"
938 ." Consider upgrading the Containers distro${extra}! Skipping further CT compat checks."
942 my $cts = eval { PVE
::API2
::LXC-
>vmlist({ node
=> $nodename }) };
944 log_warn
("Failed to retrieve information about this node's CTs - $@");
948 if (!defined($cts) || !scalar(@$cts)) {
949 log_skip
("No containers on node detected.");
953 my @running_cts = sort { $a <=> $b } grep { $_->{status
} eq 'running' } @$cts;
954 my @offline_cts = sort { $a <=> $b } grep { $_->{status
} ne 'running' } @$cts;
956 for my $ct (@running_cts) {
957 my $ctid = $ct->{vmid
};
958 my $pid = eval { PVE
::LXC
::find_lxc_pid
($ctid) };
960 log_warn
("Failed to get PID for running CT $ctid - $err");
963 my $rootdir = "/proc/$pid/root";
964 my $conf = PVE
::LXC
::Config-
>load_config($ctid);
966 my $ret = eval { $supports_cgroupv2->($conf, $rootdir, $ctid) };
968 log_warn
("Failed to get cgroup support status for CT $ctid - $err");
972 $log_problem->($ctid);
977 my $storage_cfg = PVE
::Storage
::config
();
978 for my $ct (@offline_cts) {
979 my $ctid = $ct->{vmid
};
980 my ($conf, $rootdir, $ret);
982 $conf = PVE
::LXC
::Config-
>load_config($ctid);
983 $rootdir = PVE
::LXC
::mount_all
($ctid, $storage_cfg, $conf);
984 $ret = $supports_cgroupv2->($conf, $rootdir, $ctid);
987 log_warn
("Failed to load config and mount CT $ctid - $err");
988 eval { PVE
::LXC
::umount_all
($ctid, $storage_cfg, $conf) };
992 $log_problem->($ctid);
993 eval { PVE
::LXC
::umount_all
($ctid, $storage_cfg, $conf) };
997 eval { PVE
::LXC
::umount_all
($ctid, $storage_cfg, $conf) };
1001 sub check_apt_repos
{
1002 log_info
("Checking if the suite for the Debian security repository is correct..");
1006 my $dir = '/etc/apt/sources.list.d';
1009 # TODO: check that (original) debian and Proxmox VE mirrors are present.
1011 my $check_file = sub {
1014 $file = "${dir}/${file}" if $in_dir;
1016 my $raw = eval { PVE
::Tools
::file_get_contents
($file) };
1017 return if !defined($raw);
1018 my @lines = split(/\n/, $raw);
1021 for my $line (@lines) {
1024 next if length($line) == 0; # split would result in undef then...
1026 ($line) = split(/#/, $line);
1028 next if $line !~ m/^deb[[:space:]]/; # is case sensitive
1033 # https://deb.debian.org/debian-security
1034 # http://security.debian.org/debian-security
1035 # http://security.debian.org/
1036 if ($line =~ m
|https?
://deb\
.debian\
.org
/debian-security/?\s
+(\S
*)|i
) {
1038 } elsif ($line =~ m
|https?
://security\
.debian\
.org
(?
:.*?
)\s
+(\S
*)|i
) {
1046 my $where = "in ${file}:${number}";
1048 if ($suite eq 'buster/updates') {
1049 log_info
("Make sure to change the suite of the Debian security repository " .
1050 "from 'buster/updates' to 'bullseye-security' - $where");
1051 } elsif ($suite eq 'bullseye-security') {
1052 log_pass
("already using 'bullseye-security'");
1054 log_fail
("The new suite of the Debian security repository should be " .
1055 "'bullseye-security' - $where");
1060 $check_file->("/etc/apt/sources.list");
1064 PVE
::Tools
::dir_glob_foreach
($dir, '^.*\.list$', $check_file);
1067 # only warn, it might be defined in a .sources file or in a way not catched above
1068 log_warn
("No Debian security repository detected in /etc/apt/sources.list and " .
1069 "/etc/apt/sources.list.d/*.list");
1073 sub check_time_sync
{
1074 my $unit_active = sub { return $get_systemd_unit_state->($_[0], 1) eq 'active' ?
$_[0] : undef };
1076 log_info
("Checking for supported & active NTP service..");
1077 if ($unit_active->('systemd-timesyncd.service')) {
1079 "systemd-timesyncd is not the best choice for time-keeping on servers, due to only applying"
1080 ." updates on boot.\n While not necesarry for the upgrade it's recommended to use one of:\n"
1081 ." * chrony (Default in new Proxmox VE installations)\n * ntpsec\n * openntpd\n"
1083 } elsif ($unit_active->('ntp.service')) {
1084 log_info
("Debian deprecated and removed the ntp package for Bookworm, but the system"
1085 ." will automatically migrate to the 'ntpsec' replacement package on upgrade.");
1086 } elsif (my $active_ntp = ($unit_active->('chrony.service') || $unit_active->('openntpd.service') || $unit_active->('ntpsec.service'))) {
1087 log_pass
("Detected active time synchronisation unit '$active_ntp'");
1090 "No (active) time synchronisation daemon (NTP) detected, but synchronized systems are important,"
1091 ." especially for cluster and/or ceph!"
1097 print_header
("MISCELLANEOUS CHECKS");
1098 my $ssh_config = eval { PVE
::Tools
::file_get_contents
('/root/.ssh/config') };
1099 if (defined($ssh_config)) {
1100 log_fail
("Unsupported SSH Cipher configured for root in /root/.ssh/config: $1")
1101 if $ssh_config =~ /^Ciphers .*(blowfish|arcfour|3des).*$/m;
1103 log_skip
("No SSH config file found.");
1106 log_info
("Checking common daemon services..");
1107 $log_systemd_unit_state->('pveproxy.service');
1108 $log_systemd_unit_state->('pvedaemon.service');
1109 $log_systemd_unit_state->('pvescheduler.service');
1110 $log_systemd_unit_state->('pvestatd.service');
1114 my $root_free = PVE
::Tools
::df
('/', 10);
1115 log_warn
("Less than 5 GB free space on root file system.")
1116 if defined($root_free) && $root_free->{avail
} < 5 * 1000*1000*1000;
1118 log_info
("Checking for running guests..");
1119 my $running_guests = 0;
1121 my $vms = eval { PVE
::API2
::Qemu-
>vmlist({ node
=> $nodename }) };
1122 log_warn
("Failed to retrieve information about this node's VMs - $@") if $@;
1123 $running_guests += grep { $_->{status
} eq 'running' } @$vms if defined($vms);
1125 my $cts = eval { PVE
::API2
::LXC-
>vmlist({ node
=> $nodename }) };
1126 log_warn
("Failed to retrieve information about this node's CTs - $@") if $@;
1127 $running_guests += grep { $_->{status
} eq 'running' } @$cts if defined($cts);
1129 if ($running_guests > 0) {
1130 log_warn
("$running_guests running guest(s) detected - consider migrating or stopping them.")
1132 log_pass
("no running guest detected.")
1135 log_info
("Checking if the local node's hostname '$nodename' is resolvable..");
1136 my $local_ip = eval { PVE
::Network
::get_ip_from_hostname
($nodename) };
1138 log_warn
("Failed to resolve hostname '$nodename' to IP - $@");
1140 log_info
("Checking if resolved IP is configured on local node..");
1141 my $cidr = Net
::IP
::ip_is_ipv6
($local_ip) ?
"$local_ip/128" : "$local_ip/32";
1142 my $configured_ips = PVE
::Network
::get_local_ip_from_cidr
($cidr);
1143 my $ip_count = scalar(@$configured_ips);
1145 if ($ip_count <= 0) {
1146 log_fail
("Resolved node IP '$local_ip' not configured or active for '$nodename'");
1147 } elsif ($ip_count > 1) {
1148 log_warn
("Resolved node IP '$local_ip' active on multiple ($ip_count) interfaces!");
1150 log_pass
("Resolved node IP '$local_ip' configured and active on single interface.");
1154 log_info
("Check node certificate's RSA key size");
1155 my $certs = PVE
::API2
::Certificates-
>info({ node
=> $nodename });
1157 'rsaEncryption' => {
1161 'id-ecPublicKey' => {
1167 my $certs_check_failed = 0;
1168 foreach my $cert (@$certs) {
1169 my ($type, $size, $fn) = $cert->@{qw(public-key-type public-key-bits filename)};
1171 if (!defined($type) || !defined($size)) {
1172 log_warn
("'$fn': cannot check certificate, failed to get it's type or size!");
1175 my $check = $certs_check->{$type};
1176 if (!defined($check)) {
1177 log_warn
("'$fn': certificate's public key type '$type' unknown!");
1181 if ($size < $check->{minsize
}) {
1182 log_fail
("'$fn', certificate's $check->{name} public key size is less than 2048 bit");
1183 $certs_check_failed = 1;
1185 log_pass
("Certificate '$fn' passed Debian Busters (and newer) security level for TLS connections ($size >= 2048)");
1189 check_backup_retention_settings
();
1190 check_cifs_credential_location
();
1191 check_custom_pool_roles
();
1192 check_node_and_guest_configurations
();
1197 my ($str, $color, $condition) = @_;
1198 return "". ($condition ? colored
($str, $color) : $str);
1201 __PACKAGE__-
>register_method ({
1202 name
=> 'checklist',
1203 path
=> 'checklist',
1205 description
=> 'Check (pre-/post-)upgrade conditions.',
1207 additionalProperties
=> 0,
1210 description
=> 'perform additional, expensive checks.',
1217 returns
=> { type
=> 'null' },
1221 my $kernel_cli = PVE
::Tools
::file_get_contents
('/proc/cmdline');
1222 if ($kernel_cli =~ /systemd.unified_cgroup_hierarchy=0/){
1223 $forced_legacy_cgroup = 1;
1226 check_pve_packages
();
1227 check_cluster_corosync
();
1229 check_storage_health
();
1232 if ($param->{full
}) {
1233 check_containers_cgroup_compat
();
1235 log_skip
("NOTE: Expensive checks, like CT cgroupv2 compat, not performed without '--full' parameter");
1238 print_header
("SUMMARY");
1241 $total += $_ for values %$counters;
1243 print "TOTAL: $total\n";
1244 print colored
("PASSED: $counters->{pass}\n", 'green');
1245 print "SKIPPED: $counters->{skip}\n";
1246 print colored_if
("WARNINGS: $counters->{warn}\n", 'yellow', $counters->{warn} > 0);
1247 print colored_if
("FAILURES: $counters->{fail}\n", 'red', $counters->{fail
} > 0);
1249 if ($counters->{warn} > 0 || $counters->{fail
} > 0) {
1250 my $color = $counters->{fail
} > 0 ?
'red' : 'yellow';
1251 print colored
("\nATTENTION: Please check the output for detailed information!\n", $color);
1252 print colored
("Try to solve the problems one at a time and then run this checklist tool again.\n", $color) if $counters->{fail
} > 0;
1258 our $cmddef = [ __PACKAGE__
, 'checklist', [], {}];