use PVE::Cluster;
use PVE::Storage;
use PVE::QemuServer;
+use Time::HiRes qw( usleep );
use base qw(PVE::AbstractMigrate);
}
sub finish_command_pipe {
- my ($self, $cmdpipe) = @_;
+ my ($self, $cmdpipe, $timeout) = @_;
my $writer = $cmdpipe->{writer};
my $reader = $cmdpipe->{reader};
my $cpid = $cmdpipe->{pid};
- kill(15, $cpid) if kill(0, $cpid);
-
- waitpid($cpid, 0);
-}
-
-sub run_with_timeout {
- my ($timeout, $code, @param) = @_;
-
- die "got timeout\n" if $timeout <= 0;
-
- my $prev_alarm;
-
- my $sigcount = 0;
-
- my $res;
-
- eval {
- local $SIG{ALRM} = sub { $sigcount++; die "got timeout\n"; };
- local $SIG{PIPE} = sub { $sigcount++; die "broken pipe\n" };
- local $SIG{__DIE__}; # see SA bug 4631
-
- $prev_alarm = alarm($timeout);
-
- $res = &$code(@param);
-
- alarm(0); # avoid race conditions
- };
-
- my $err = $@;
-
- alarm($prev_alarm) if defined($prev_alarm);
+ if ($timeout) {
+ for (my $i = 0; $i < $timeout; $i++) {
+ return if !PVE::ProcFSTools::check_process_running($cpid);
+ sleep(1);
+ }
+ }
- die "unknown error" if $sigcount && !$err; # seems to happen sometimes
+ $self->log('info', "ssh tunnel still running - terminating now with SIGTERM\n");
+ kill(15, $cpid);
- die $err if $err;
+ # wait again
+ for (my $i = 0; $i < 10; $i++) {
+ return if !PVE::ProcFSTools::check_process_running($cpid);
+ sleep(1);
+ }
- return $res;
+ $self->log('info', "ssh tunnel still running - terminating now with SIGKILL\n");
+ kill 9, $cpid;
+ sleep 1;
}
sub fork_tunnel {
my $helo;
eval {
- run_with_timeout(60, sub { $helo = <$reader>; });
+ PVE::Tools::run_with_timeout(60, sub { $helo = <$reader>; });
die "no reply\n" if !$helo;
die "no quorum on target node\n" if $helo =~ m/^no quorum$/;
die "got strange reply from mtunnel ('$helo')\n"
my $writer = $tunnel->{writer};
eval {
- run_with_timeout(30, sub {
+ PVE::Tools::run_with_timeout(30, sub {
print $writer "quit\n";
$writer->flush();
});
};
my $err = $@;
- $self->finish_command_pipe($tunnel);
+ $self->finish_command_pipe($tunnel, 30);
die $err if $err;
}
sub lock_vm {
my ($self, $vmid, $code, @param) = @_;
-
+
return PVE::QemuServer::lock_config($vmid, $code, @param);
}
my $volhash = {};
my $cdromhash = {};
- # get list from PVE::Storage (for unused volumes)
- my $dl = PVE::Storage::vdisk_list($self->{storecfg}, undef, $vmid);
- PVE::Storage::foreach_volid($dl, sub {
- my ($volid, $sid, $volname) = @_;
+ my @sids = PVE::Storage::storage_ids($self->{storecfg});
+ foreach my $storeid (@sids) {
+ my $scfg = PVE::Storage::storage_config($self->{storecfg}, $storeid);
+ next if $scfg->{shared};
+ next if !PVE::Storage::storage_check_enabled($self->{storecfg}, $storeid, undef, 1);
- # check if storage is available on both nodes
- my $scfg = PVE::Storage::storage_check_node($self->{storecfg}, $sid);
- PVE::Storage::storage_check_node($self->{storecfg}, $sid, $self->{node});
+ # get list from PVE::Storage (for unused volumes)
+ my $dl = PVE::Storage::vdisk_list($self->{storecfg}, $storeid, $vmid);
+ PVE::Storage::foreach_volid($dl, sub {
+ my ($volid, $sid, $volname) = @_;
- return if $scfg->{shared};
+ # check if storage is available on target node
+ PVE::Storage::storage_check_node($self->{storecfg}, $sid, $self->{node});
- $volhash->{$volid} = 1;
- });
+ $volhash->{$volid} = 1;
+ });
+ }
# and add used,owned/non-shared disks (just to be sure we have all)
my $conf = $self->{vmconf};
# set migrate lock in config file
- PVE::QemuServer::change_config_nolock($vmid, { lock => 'migrate' }, {}, 1);
+ $conf->{lock} = 'migrate';
+ PVE::QemuServer::update_config_nolock($vmid, $conf, 1);
sync_disks($self, $vmid);
- # move config to remote node
- my $conffile = PVE::QemuServer::config_file($vmid);
- my $newconffile = PVE::QemuServer::config_file($vmid, $self->{node});
-
- die "Failed to move config to node '$self->{node}' - rename failed: $!\n"
- if !rename($conffile, $newconffile);
};
sub phase1_cleanup {
$self->log('info', "aborting phase 1 - cleanup resources");
- my $unset = { lock => 1 };
- eval { PVE::QemuServer::change_config_nolock($vmid, {}, $unset, 1) };
+ my $conf = $self->{vmconf};
+ delete $conf->{lock};
+ eval { PVE::QemuServer::update_config_nolock($vmid, $conf, 1) };
if (my $err = $@) {
$self->log('err', $err);
}
-
+
if ($self->{volumes}) {
foreach my $volid (@{$self->{volumes}}) {
$self->log('err', "found stale volume copy '$volid' on node '$self->{node}'");
my $rport;
+ my $nodename = PVE::INotify::nodename();
+
## start on remote node
- my $cmd = [@{$self->{rem_ssh}}, 'qm', 'start',
- $vmid, '--stateuri', 'tcp', '--skiplock'];
+ my $cmd = [@{$self->{rem_ssh}}, 'qm', 'start',
+ $vmid, '--stateuri', 'tcp', '--skiplock', '--migratedfrom', $nodename];
PVE::Tools::run_command($cmd, outfunc => sub {
my $line = shift;
my $start = time();
- my $merr = PVE::QemuServer::vm_monitor_command($vmid, "migrate -d \"tcp:localhost:$lport\"", 1);
+ my $capabilities = {};
+ $capabilities->{capability} = "xbzrle";
+ $capabilities->{state} = JSON::true;
+
+ eval {
+ PVE::QemuServer::vm_mon_cmd_nocheck($vmid, "migrate-set-capabilities", capabilities => [$capabilities]);
+ };
- my $lstat = '';
+ #set cachesize 10% of the total memory
+ my $cachesize = int($conf->{memory}*1048576/10);
+ eval {
+ PVE::QemuServer::vm_mon_cmd_nocheck($vmid, "migrate-set-cache-size", value => $cachesize);
+ };
+
+ eval {
+ PVE::QemuServer::vm_mon_cmd_nocheck($vmid, "migrate", uri => "tcp:localhost:$lport");
+ };
+ my $merr = $@;
+
+ my $lstat = 0;
+ my $usleep = 2000000;
+ my $i = 0;
+ my $err_count = 0;
while (1) {
- sleep (2);
- my $stat = PVE::QemuServer::vm_monitor_command($vmid, "info migrate", 1);
- if ($stat =~ m/^Migration status: (active|completed|failed|cancelled)$/im) {
- $merr = undef;
- my $ms = $1;
-
- if ($stat ne $lstat) {
- if ($ms eq 'active') {
- my ($trans, $rem, $total) = (0, 0, 0);
- $trans = $1 if $stat =~ m/^transferred ram: (\d+) kbytes$/im;
- $rem = $1 if $stat =~ m/^remaining ram: (\d+) kbytes$/im;
- $total = $1 if $stat =~ m/^total ram: (\d+) kbytes$/im;
-
- $self->log('info', "migration status: $ms (transferred ${trans}KB, " .
- "remaining ${rem}KB), total ${total}KB)");
- } else {
- $self->log('info', "migration status: $ms");
- }
+ $i++;
+ my $avglstat = $lstat/$i if $lstat;
+
+ usleep($usleep);
+ my $stat;
+ eval {
+ $stat = PVE::QemuServer::vm_mon_cmd_nocheck($vmid, "query-migrate");
+ };
+ if (my $err = $@) {
+ $err_count++;
+ warn "query migrate failed: $err\n";
+ if ($err_count <= 5) {
+ usleep(1000000);
+ next;
}
-
- if ($ms eq 'completed') {
+ die "too many query migrate failures - aborting\n";
+ }
+ if ($stat->{status} =~ m/^(active|completed|failed|cancelled)$/im) {
+ $merr = undef;
+ $err_count = 0;
+ if ($stat->{status} eq 'completed') {
my $delay = time() - $start;
if ($delay > 0) {
my $mbps = sprintf "%.2f", $conf->{memory}/$delay;
$self->log('info', "migration speed: $mbps MB/s");
}
}
-
- if ($ms eq 'failed' || $ms eq 'cancelled') {
+
+ if ($stat->{status} eq 'failed' || $stat->{status} eq 'cancelled') {
die "aborting\n"
}
- last if $ms ne 'active';
+ if ($stat->{status} ne 'active') {
+ $self->log('info', "migration status: $stat->{status}");
+ last;
+ }
+
+ if ($stat->{ram}->{transferred} ne $lstat) {
+ my $trans = $stat->{ram}->{transferred} || 0;
+ my $rem = $stat->{ram}->{remaining} || 0;
+ my $total = $stat->{ram}->{total} || 0;
+ my $xbzrlecachesize = $stat->{"xbzrle-cache"}->{"cache-size"} || 0;
+ my $xbzrlebytes = $stat->{"xbzrle-cache"}->{"bytes"} || 0;
+ my $xbzrlepages = $stat->{"xbzrle-cache"}->{"pages"} || 0;
+ my $xbzrlecachemiss = $stat->{"xbzrle-cache"}->{"cache-miss"} || 0;
+ my $xbzrleoverflow = $stat->{"xbzrle-cache"}->{"overflow"} || 0;
+ #reduce sleep if remainig memory if lower than the everage transfert
+ $usleep = 300000 if $rem < $avglstat;
+
+ $self->log('info', "migration status: $stat->{status} (transferred ${trans}, " .
+ "remaining ${rem}), total ${total})");
+
+ $self->log('info', "migration xbzrle cachesize: ${xbzrlecachesize} transferred ${xbzrlebytes} pages ${xbzrlepages} cachemiss ${xbzrlecachemiss} overflow ${xbzrleoverflow}");
+ }
+
+ $lstat = $stat->{ram}->{transferred};
+
} else {
die $merr if $merr;
- die "unable to parse migration status '$stat' - aborting\n";
+ die "unable to parse migration status '$stat->{status}' - aborting\n";
}
- $lstat = $stat;
- };
+ }
+}
+
+sub phase2_cleanup {
+ my ($self, $vmid, $err) = @_;
+
+ return if !$self->{errors};
+ $self->{phase2errors} = 1;
+
+ $self->log('info', "aborting phase 2 - cleanup resources");
+
+ my $conf = $self->{vmconf};
+ delete $conf->{lock};
+ eval { PVE::QemuServer::update_config_nolock($vmid, $conf, 1) };
+ if (my $err = $@) {
+ $self->log('err', $err);
+ }
+
+ # cleanup ressources on target host
+ my $nodename = PVE::INotify::nodename();
+
+ my $cmd = [@{$self->{rem_ssh}}, 'qm', 'stop', $vmid, '--skiplock', '--migratedfrom', $nodename];
+ eval{ PVE::Tools::run_command($cmd, outfunc => sub {}, errfunc => sub {}) };
+ if (my $err = $@) {
+ $self->log('err', $err);
+ $self->{errors} = 1;
+ }
}
sub phase3 {
my ($self, $vmid) = @_;
-
+
my $volids = $self->{volumes};
+ return if $self->{phase2errors};
# destroy local copies
foreach my $volid (@$volids) {
last if $err =~ /^interrupted by signal$/;
}
}
+}
+sub phase3_cleanup {
+ my ($self, $vmid, $err) = @_;
+
+ my $conf = $self->{vmconf};
+ return if $self->{phase2errors};
+
+ # move config to remote node
+ my $conffile = PVE::QemuServer::config_file($vmid);
+ my $newconffile = PVE::QemuServer::config_file($vmid, $self->{node});
+
+ die "Failed to move config to node '$self->{node}' - rename failed: $!\n"
+ if !rename($conffile, $newconffile);
+
+ # now that config file is move, we can resume vm on target if livemigrate
if ($self->{tunnel}) {
- eval { finish_tunnel($self, $self->{tunnel}); };
+ my $cmd = [@{$self->{rem_ssh}}, 'qm', 'resume', $vmid, '--skiplock'];
+ eval{ PVE::Tools::run_command($cmd, outfunc => sub {}, errfunc => sub {}) };
if (my $err = $@) {
$self->log('err', $err);
$self->{errors} = 1;
}
}
-}
-
-sub phase3_cleanup {
- my ($self, $vmid, $err) = @_;
-
- my $conf = $self->{vmconf};
# always stop local VM
eval { PVE::QemuServer::vm_stop($self->{storecfg}, $vmid, 1, 1); };
$self->{errors} = 1;
}
+ if ($self->{tunnel}) {
+ eval { finish_tunnel($self, $self->{tunnel}); };
+ if (my $err = $@) {
+ $self->log('err', $err);
+ $self->{errors} = 1;
+ }
+ }
+
# always deactivate volumes - avoid lvm LVs to be active on several nodes
eval {
my $vollist = PVE::QemuServer::get_vm_volumes($conf);