marsadm: log all messages to syslog

This commit is contained in:
Thomas Schoebel-Theuer 2012-12-25 22:41:38 +01:00
parent 70d409b200
commit c6d22e7553
1 changed files with 197 additions and 161 deletions

View File

@ -5,14 +5,48 @@ use strict;
use English;
use warnings;
my $Id = "267700a8a877ad79bc9b86f17891c9d7bf6fa766";
umask 0077;
##################################################################
# messaging
my $warn = "";
my $notify = "";
sub lprint {
my ($text) = @_;
print "$warn$text";
if ($notify) {
system("/usr/bin/logger -t marsadm \"$warn$notify $text\"");
}
}
sub ldie {
my ($text) = @_;
$warn = "DYING: ";
lprint $text;
exit -1;
}
sub lwarn {
my ($text) = @_;
my $oldwarn = $warn;
$warn = "WARNING: ";
lprint $text;
$warn = $oldwarn;
}
##################################################################
# global variables
my $Id = '$Id$';
my $mars = "/mars";
my $host = `uname -n` or die "cannot determine my network node name\n";
my $host = `uname -n` or ldie "cannot determine my network node name\n";
chomp $host;
my $force = 0;
my $ip = _get_ip() or die "cannot determine my IP address\n";
umask 0077;
my $ip = _get_ip() or ldie "cannot determine my IP address\n";
##################################################################
@ -20,8 +54,8 @@ umask 0077;
sub check_id {
my $str = shift;
die "identifier '$str' has disallowed characters" unless $str =~ m/^[A-Za-z_][-A-Za-z0-9_]*$/;
die "identifier '$str' is too long (only 16 chars allowed)" if length($str) > 16;
ldie "identifier '$str' has disallowed characters" unless $str =~ m/^[A-Za-z_][-A-Za-z0-9_]*$/;
ldie "identifier '$str' is too long (only 16 chars allowed)" if length($str) > 16;
}
##################################################################
@ -53,9 +87,9 @@ sub check_res {
}
}
}
die "resource '$res' does not exist ($count replacements found)\n" unless $count == 1 and $found;
ldie "resource '$res' does not exist ($count replacements found)\n" unless $count == 1 and $found;
$found =~ s:^.*/resource-(.*)/.*$:$1:;
warn "substituting bad resource name '$res' by uniquely matching resource name '$found'\n";
lwarn "substituting bad resource name '$res' by uniquely matching resource name '$found'\n";
$res = $found;
}
return $res;
@ -63,7 +97,7 @@ sub check_res {
sub check_res_member {
my $res = shift;
die "sorry, I have not yet joined to resource '$res'\n" unless -e "$mars/resource-$res/data-$host";
ldie "sorry, I have not yet joined to resource '$res'\n" unless -e "$mars/resource-$res/data-$host";
}
sub check_sync_finished {
@ -71,24 +105,24 @@ sub check_sync_finished {
my $lnk = "$mars/resource-$res/syncstatus-$host";
if(lstat($lnk)) {
my $syncstatus = readlink($lnk);
my $size = readlink("$mars/resource-$res/size") or die "cannot read size\n";
die "sync has not yet finished, only $syncstatus / $size bytes transferred\n" unless $syncstatus >= $size;
my $size = readlink("$mars/resource-$res/size") or ldie "cannot read size\n";
ldie "sync has not yet finished, only $syncstatus / $size bytes transferred\n" unless $syncstatus >= $size;
}
print "OK, it seems that sync has finished on $host.\n";
lprint "OK, it seems that sync has finished on $host.\n";
}
sub check_primary {
my ($cmd, $res) = @_;
my $pri = "$mars/resource-$res/primary";
my $old = readlink($pri) or die "cannot determine current primary\n";
die "for operation '$cmd' I need to be primary\n" unless $old eq $host;
my $old = readlink($pri) or ldie "cannot determine current primary\n";
ldie "for operation '$cmd' I need to be primary\n" unless $old eq $host;
}
sub check_not_primary {
my ($cmd, $res) = @_;
my $pri = "$mars/resource-$res/primary";
my $old = readlink($pri) or die "cannot determine current primary\n";
die "operation '$cmd' cannot be executed on primary\n" if $old eq $host;
my $old = readlink($pri) or ldie "cannot determine current primary\n";
ldie "operation '$cmd' cannot be executed on primary\n" if $old eq $host;
}
sub check_primary_gone {
@ -101,7 +135,7 @@ sub check_primary_gone {
$found++ if $val;
}
last if !$found;
print "waiting for $found other primary host(s) to disappear....\n";
lprint "waiting for $found other primary host(s) to disappear....\n";
sleep(5);
}
}
@ -110,10 +144,10 @@ sub check_todo {
my ($res, $key, $val, $wait) = @_;
for(;;) {
my $path = "$mars/resource-$res/todo-$host/$key";
my $link = readlink($path) or die "cannot read symlink '$path'\n";
my $link = readlink($path) or ldie "cannot read symlink '$path'\n";
last if $link == $val;
die "$path must have value $val\n" if $wait;
print "waiting until $key reaches the value $val....\n";
ldie "$path must have value $val\n" if $wait;
lprint "waiting until $key reaches the value $val....\n";
sleep(5);
}
}
@ -122,10 +156,10 @@ sub check_status {
my ($res, $key, $val, $wait) = @_;
for(;;) {
my $path = "$mars/resource-$res/actual-$host/$key";
my $link = readlink($path) or die "cannot read symlink '$path'\n";
my $link = readlink($path) or ldie "cannot read symlink '$path'\n";
last if $link == $val;
die "$path must have value $val\n" if $wait;
print "waiting until $key reaches the value $val....\n";
ldie "$path must have value $val\n" if $wait;
lprint "waiting until $key reaches the value $val....\n";
sleep(5);
}
}
@ -137,7 +171,7 @@ sub _check_mtime {
return 0;
}
my $res = ($mt < time() - $age);
#print "XXX '$path' $res\n";
#lprint "XXX '$path' $res\n";
return $res;
}
@ -157,11 +191,11 @@ sub _get_minmax {
my ($res, $glob, $take_symlink) = @_;
my $min = -1;
my $max = -1;
my @paths = glob($glob) or die "cannot find '$glob'\n";
my @paths = glob($glob) or ldie "cannot find '$glob'\n";
foreach my $path (@paths) {
my $nr = $path;
if($take_symlink) {
$nr = readlink($path) or die "cannot read symlink '$path'\n";
$nr = readlink($path) or ldie "cannot read symlink '$path'\n";
}
$nr =~ s:^.*[a-z]+-([0-9]+)(-[^/]*)?$:$1:;
$min = $nr if ($nr < $min || $min < 0);
@ -196,12 +230,12 @@ sub check_splitbrain {
my ($res, $host, $sequence) = @_;
if($sequence < 0) {
my $pri = "$mars/resource-$res/primary";
my $old = readlink($pri) or die "cannot determine current primary\n";
my $old = readlink($pri) or ldie "cannot determine current primary\n";
_primary_res($res, "(none)", $pri, $old) unless $old eq "(none)";
_trigger();
sleep(5);
while(!_check_all_mtimes("$mars/resource-$res/{log,version,replay}-*", 60)) {
print "resource directory $res not stable, waiting....\n";
lprint "resource directory $res not stable, waiting....\n";
sleep(5);
}
while(1) {
@ -209,15 +243,15 @@ sub check_splitbrain {
my ($min_ver, $max_ver) = get_minmax_versions($res);
my ($min_rep, $max_rep) = get_minmax_replays($res);
if($min_ver > $min_log || $max_ver < $max_log) {
print "some version links are missing...\n";
lprint "some version links are missing...\n";
sleep(10);
next;
}
if($max_log >= $max_rep) {
print "resource $res: logfile $max_log is present.\n";
lprint "resource $res: logfile $max_log is present.\n";
last;
}
print "resource $res: logfile $max_log is not yet transferred (need $max_rep), waiting....\n";
lprint "resource $res: logfile $max_log is not yet transferred (need $max_rep), waiting....\n";
sleep(10);
}
} # $sequence < 0
@ -226,8 +260,8 @@ sub check_splitbrain {
my @links = glob($glob);
if(!@links) {
@links = glob("$mars/resource-$res/version-[0-9]*-*");
die "no version information available\n" unless @links;
print "assuming that I am primary for the first time\n";
ldie "no version information available\n" unless @links;
lprint "assuming that I am primary for the first time\n";
return;
}
@ -242,11 +276,11 @@ sub check_splitbrain {
my $fromhost = $link;
$fromhost =~ s:^.*version-[0-9]*-(.*)$:$1:;
my $version = readlink($link) or die "cannot read symlink '$link'\n";
my $version = readlink($link) or ldie "cannot read symlink '$link'\n";
my $otherhost = $version;
$otherhost =~ s:^[^,]*,([^,]*),.*$:$1:;
my $otherlink = sprintf("$mars/resource-$res/version-%09d-$otherhost", $nr);
my $otherversion = readlink($otherlink) or die "cannot read symlink '$otherlink'\n";
my $otherversion = readlink($otherlink) or ldie "cannot read symlink '$otherlink'\n";
# ignore foreign mismatches
if($host) {
@ -257,7 +291,7 @@ sub check_splitbrain {
next if $otherhost eq $fromhost;
# final check
die "splitbrain at sequence $nr detected\n" unless $version eq $otherversion;
ldie "splitbrain at sequence $nr detected\n" unless $version eq $otherversion;
}
}
@ -298,23 +332,23 @@ sub _switch {
my $old = readlink($path);
if($old && $old eq $src) {
print "${cmd} on resource $res is already activated\n" if $cmd;
lprint "${cmd} on resource $res is already activated\n" if $cmd;
return;
}
my $tmp = $path;
$tmp =~ s/\/([^\/]+)$/.tmp.$1/;
symlink($src, $tmp) or die "cannot create switch symlink\n";
rename($tmp, $path) or die "cannot rename switch symlink\n";
print "successfully started ${cmd} on resource $res\n" if $cmd;
symlink($src, $tmp) or ldie "cannot create switch symlink\n";
rename($tmp, $path) or ldie "cannot rename switch symlink\n";
lprint "successfully started ${cmd} on resource $res\n" if $cmd;
}
sub _writable {
my ($path, $on) = @_;
my $oldmode = (lstat $path)[2] & 0700;
my $newmode = $on ? $oldmode | 0200 : $oldmode & ~0200;
print "chmod '$path' $oldmode $newmode";
chmod($newmode, $path) == 1 or die "cannot chmod '$path'\n";
lprint "chmod '$path' $oldmode $newmode";
chmod($newmode, $path) == 1 or ldie "cannot chmod '$path'\n";
}
sub _get_ip {
@ -338,13 +372,13 @@ sub _fake_versionlink {
$prevlink = shift @test;
}
if ($prevlink) {
print "creating faked version symlink...\n";
lprint "creating faked version symlink...\n";
my $myversion = sprintf("$basedir/version-%09d-$host", $prev);
system("rm -f $myversion.tmp");
symlink($prevlink, "$myversion.tmp") or die "cannot create faked version symlink '$myversion'\n";
symlink($prevlink, "$myversion.tmp") or ldie "cannot create faked version symlink '$myversion'\n";
system("mv $myversion.tmp $myversion");
} else {
warn "cannot read symlink '$prevversion' -- cannot create fake\n";
lwarn "cannot read symlink '$prevversion' -- cannot create fake\n";
}
}
}
@ -366,7 +400,7 @@ sub _set_replaylink {
# copy pervious version symlink
_fake_versionlink($basedir, $replay, $primary);
# create replay symlink
symlink($replay, "$replaylink.tmp") or die "cannot create replay status\n";
symlink($replay, "$replaylink.tmp") or ldie "cannot create replay status\n";
system("mv $replaylink.tmp $replaylink");
}
@ -376,24 +410,24 @@ sub _set_replaylink {
sub ignore_cmd {
my ($cmd, $res) = @_;
print "ignoring command '$cmd' on resource '$res'\n";
lprint "ignoring command '$cmd' on resource '$res'\n";
exit(0);
}
sub senseless_cmd {
my ($cmd, $res) = @_;
print "command '$cmd' makes no sense with MARS (ignoring)\n";
lprint "command '$cmd' makes no sense with MARS (ignoring)\n";
exit(0);
}
sub forbidden_cmd {
my ($cmd, $res) = @_;
die "command '$cmd' cannot be used with MARS (it is impossible to carry out uniquely and could therefore lead to a disaster)\n";
ldie "command '$cmd' cannot be used with MARS (it is impossible to carry out uniquely and could therefore lead to a disaster)\n";
}
sub nyi_cmd {
my ($cmd, $res) = @_;
die "command '$cmd' is not yet implemented\n";
ldie "command '$cmd' is not yet implemented\n";
}
sub is_module_loaded {
@ -417,30 +451,30 @@ sub _create_cluster {
sub create_cluster {
my ($cmd, $peer) = @_;
die "cluster is already created\n" if !$force && -d "$mars/ips";
die "mars module is loaded, please unload first\n" if is_module_loaded();
ldie "cluster is already created\n" if !$force && -d "$mars/ips";
ldie "mars module is loaded, please unload first\n" if is_module_loaded();
_create_cluster(@_);
}
sub join_cluster {
my ($cmd, $peer) = @_;
if(glob("$mars/resource-*") or glob("$mars/ips/*")) {
die "Sorry, some resources already exist!\nThis is dangerous!\nIf you are sure that no resource clash is possible, re-invoke this command with '--force' option\n" unless $force;
ldie "Sorry, some resources already exist!\nThis is dangerous!\nIf you are sure that no resource clash is possible, re-invoke this command with '--force' option\n" unless $force;
}
die "mars module is loaded, please unload first\n" if is_module_loaded();
print "joining cluster via rsync (peer='$peer')\n";
ldie "mars module is loaded, please unload first\n" if is_module_loaded();
lprint "joining cluster via rsync (peer='$peer')\n";
# check connection
system("ssh $peer uname -a") == 0 or die "oops, no connection to $peer ...\n";
system("ssh $peer uname -a") == 0 or ldie "oops, no connection to $peer ...\n";
_create_cluster(@_);
system("rsync --recursive --links -v $peer:$mars/ips/ $mars/ips/") == 0 or die "oops\n";
system("rsync --recursive --links -v $peer:$mars/ips/ $mars/ips/") == 0 or ldie "oops\n";
symlink($ip, "$mars/ips/ip-$host");
system("rsync --recursive --links -v $mars/ips/ $peer:$mars/ips/") == 0 or die "oops\n";
system("rsync --recursive --links -v $mars/ips/ $peer:$mars/ips/") == 0 or ldie "oops\n";
}
sub leave_cluster {
my ($cmd) = @_;
my $check = "/mars/resource-*/*-$host";
die "I am member of some resources\n" if glob($check) && !$force;
ldie "I am member of some resources\n" if glob($check) && !$force;
_create_delete("$mars/ips/$host");
}
@ -448,30 +482,30 @@ sub create_res {
my ($cmd, $res, $dev, $appear) = @_;
my $create = ($cmd eq "create-resource");
die "undefined device or size argument\n" unless $dev;
ldie "undefined device or size argument\n" unless $dev;
$appear = $res if !$appear;
check_id($appear) if $appear;
if($create) {
die "resource '$res' already exists\n" if -d "$mars/resource-$res";
print "creating new resource '$res'\n";
ldie "resource '$res' already exists\n" if -d "$mars/resource-$res";
lprint "creating new resource '$res'\n";
} else {
die "resource '$res' has been already joined -- this is dangerous!\n" if -e "$mars/resource-$res/connect-$host";
print "joining to existing resource '$res'\n";
ldie "resource '$res' has been already joined -- this is dangerous!\n" if -e "$mars/resource-$res/connect-$host";
lprint "joining to existing resource '$res'\n";
}
my $size = get_size($dev);
if($size > 0) {
$dev = "";
} else {
die "block device '$dev' does not exist\n" unless -b $dev;
die "block device '$dev' must be an absolute path starting with '/'\n" unless $dev =~ m/^\//;
ldie "block device '$dev' does not exist\n" unless -b $dev;
ldie "block device '$dev' must be an absolute path starting with '/'\n" unless $dev =~ m/^\//;
use Fcntl 'SEEK_END';
open(TEST, "<$dev") or die "cannot open device for reading\n";
open(TEST, "<$dev") or ldie "cannot open device for reading\n";
$size = sysseek(TEST, 0, SEEK_END);
close(TEST);
print "device size = $size bytes\n";
die "implausible size $size" unless $size > 0;
lprint "device size = $size bytes\n";
ldie "implausible size $size" unless $size > 0;
}
my $tmp = "$mars/.tmp.$res";
@ -480,48 +514,48 @@ sub create_res {
if($create) {
_create_cluster(@_);
system("rm -rf $tmp");
system("mkdir $tmp") == 0 or die "could not create resource '$res'\n";
symlink($size, "$tmp/size") or die "cannot create size indicator symlink\n";
system("mkdir $tmp") == 0 or ldie "could not create resource '$res'\n";
symlink($size, "$tmp/size") or ldie "cannot create size indicator symlink\n";
} else {
$tmp = "$mars/resource-$res";
die "resource '$res' does not exist\n" unless -d $tmp;
$primary = readlink("$tmp/primary") or die "cannot determine primary\n";
ldie "resource '$res' does not exist\n" unless -d $tmp;
$primary = readlink("$tmp/primary") or ldie "cannot determine primary\n";
if($primary eq "(none)") {
my @list = glob("$tmp/replay-*") or die "cannot find any candidate for primary\n";
my $first = pop @list or die "bad glob list\n";
$primary = readlink($first) or die "cannot determine peer\n";
my @list = glob("$tmp/replay-*") or ldie "cannot find any candidate for primary\n";
my $first = pop @list or ldie "bad glob list\n";
$primary = readlink($first) or ldie "cannot determine peer\n";
$primary =~ s/^log-[0-9]+-(.*),.*,.*$/$1/;
print "using '$primary' as primary\n";
lprint "using '$primary' as primary\n";
}
die "resource '$res' is already joined\n" if -e "$tmp/data-$host";
die "my ip '$ip' is not registered -- please run 'join-cluster' first\n" unless -l "$mars/ips/ip-$host";
my $oldsize = readlink("$tmp/size") or die "cannot determine old size\n";
ldie "resource '$res' is already joined\n" if -e "$tmp/data-$host";
ldie "my ip '$ip' is not registered -- please run 'join-cluster' first\n" unless -l "$mars/ips/ip-$host";
my $oldsize = readlink("$tmp/size") or ldie "cannot determine old size\n";
if($size < $oldsize) {
print "adjusting size to $oldsize\n";
lprint "adjusting size to $oldsize\n";
$size = $oldsize;
}
die "sizes differ: real size = $oldsize, but requested size = $size\n" unless $oldsize == $size;
$replay = readlink("$tmp/replay-$primary") or die "cannot read replay status of primary '$primary'\n";
ldie "sizes differ: real size = $oldsize, but requested size = $size\n" unless $oldsize == $size;
$replay = readlink("$tmp/replay-$primary") or ldie "cannot read replay status of primary '$primary'\n";
$replay =~ s/,[0-9]+,[0-9]+$/,0,0/;
}
my $file = "$tmp/data-$host";
if(!$dev) {
print "creating sparse file '$file' with size $size\n";
open(OUT, ">$file") or die "could not open '$file'\n";
lprint "creating sparse file '$file' with size $size\n";
open(OUT, ">$file") or ldie "could not open '$file'\n";
use Fcntl 'SEEK_SET';
sysseek(OUT, $size-1, SEEK_SET) == $size-1 or die "could not seek\n";
syswrite(OUT, '\0', 1) == 1 or die "cannot init sparse file\n";
sysseek(OUT, $size-1, SEEK_SET) == $size-1 or ldie "could not seek\n";
syswrite(OUT, '\0', 1) == 1 or ldie "cannot init sparse file\n";
close OUT;
} else {
print "using existing device '$dev'\n";
symlink($dev, $file) or die "cannot create device symlink\n";
lprint "using existing device '$dev'\n";
symlink($dev, $file) or ldie "cannot create device symlink\n";
}
if($appear) {
# TODO: check for uniqeness of $appear
print "resource '$res' will appear as local device '/dev/mars/$appear'\n";
lprint "resource '$res' will appear as local device '/dev/mars/$appear'\n";
system("rm -f $tmp/device-$host");
symlink($appear, "$tmp/device-$host") or die "cannot create symlink for local device appearance\n";
symlink($appear, "$tmp/device-$host") or ldie "cannot create symlink for local device appearance\n";
}
mkdir("$tmp/userspace") unless -d "$tmp/userspace";
@ -537,19 +571,19 @@ sub create_res {
system("rm -f $tmp/syncstatus-$host");
if($create) {
symlink($host, "$tmp/primary") or die "cannot create primary symlink\n";
symlink($size, "$tmp/syncstatus-$host") or die "cannot create primary syncstatus\n";
symlink("log-000000001-$host,0,0", "$tmp/replay-$host") or die "cannot create replay status\n";
symlink($host, "$tmp/primary") or ldie "cannot create primary symlink\n";
symlink($size, "$tmp/syncstatus-$host") or ldie "cannot create primary syncstatus\n";
symlink("log-000000001-$host,0,0", "$tmp/replay-$host") or ldie "cannot create replay status\n";
system("touch $tmp/log-000000001-$host");
rename($tmp, "$mars/resource-$res") or die "cannot finalize resource '$res'\n";
print "successfully created resource '$res'\n";
rename($tmp, "$mars/resource-$res") or ldie "cannot finalize resource '$res'\n";
lprint "successfully created resource '$res'\n";
} else {
_set_replaylink($tmp, $replay, $primary);
symlink("0", "$tmp/syncstatus-$host") or die "cannot start initial sync\n";
symlink("0", "$tmp/syncstatus-$host") or ldie "cannot start initial sync\n";
system("rm -f $tmp/connect-$host");
symlink($primary, "$tmp/connect-$host") or die "cannot create peer connect symlink\n";
symlink($primary, "$tmp/connect-$host") or ldie "cannot create peer connect symlink\n";
symlink($host, "$tmp/connect-$primary") unless -l "$tmp/connect-$primary";
print "successfully joined resource '$res'\n";
lprint "successfully joined resource '$res'\n";
}
}
@ -558,24 +592,24 @@ sub leave_res {
check_not_primary(@_);
foreach my $tmp (glob("$mars/resource-$res/todo-$host/*")) {
my $status = readlink($tmp);
die "cannot read symlink '$tmp'\n" unless defined($status);
die "switch '$tmp' is not off\n" if $status;
ldie "cannot read symlink '$tmp'\n" unless defined($status);
ldie "switch '$tmp' is not off\n" if $status;
}
foreach my $tmp (glob("$mars/resource-$res/actual-$host/*")) {
my $status = readlink($tmp);
die "cannot read symlink '$tmp'\n" unless defined($status);
die "running status '$tmp' is not off\n" if $status;
ldie "cannot read symlink '$tmp'\n" unless defined($status);
ldie "running status '$tmp' is not off\n" if $status;
}
my $peerlink = "$mars/resource-$res/connect-$host";
my $peer = readlink($peerlink) or die "cannot read symlink '$peerlink'\n";
my $peer = readlink($peerlink) or ldie "cannot read symlink '$peerlink'\n";
foreach my $tmp (glob("$mars/resource-$res/connect-*")) {
next if $tmp eq $peerlink;
my $target = readlink($tmp) or die "cannot read symlink '$tmp'\n";
my $target = readlink($tmp) or ldie "cannot read symlink '$tmp'\n";
next unless $target eq $host;
print "changing '$tmp' from '$host' to '$peer'\n";
lprint "changing '$tmp' from '$host' to '$peer'\n";
unlink("$tmp.new");
symlink($peer, "$tmp.new") or die "cannot create symlink '$tmp.new'\n";
rename("$tmp.new", $tmp) or die "cannot create symlink '$tmp'\n";
symlink($peer, "$tmp.new") or ldie "cannot create symlink '$tmp.new'\n";
rename("$tmp.new", $tmp) or ldie "cannot create symlink '$tmp'\n";
}
unlink($peerlink);
}
@ -583,31 +617,31 @@ sub leave_res {
sub logrotate_res {
my ($cmd, $res) = @_;
check_primary(@_);
my @paths = glob("$mars/resource-$res/log-*-$host") or die "cannot find any logfiles\n";
my @paths = glob("$mars/resource-$res/log-*-$host") or ldie "cannot find any logfiles\n";
@paths = sort(@paths);
my $last = pop(@paths);
if(-z $last) {
print "an empty logfile '$last' already exists, nothing to do.\n";
lprint "an empty logfile '$last' already exists, nothing to do.\n";
return;
}
my $nr = $last;
$nr =~ s/^.*log-([0-9]+)-.+$/$1/;
my $next = sprintf("$mars/resource-$res/log-%09d-$host", $nr + 1);
die "logfile '$next' already exists\n" if -e $next;
ldie "logfile '$next' already exists\n" if -e $next;
system("touch $next");
}
sub _allowed_logdelete {
my ($cmd, $res) = @_;
my $min = -1;
my @paths = glob("$mars/resource-$res/replay-*") or die "cannot find any replay symlinks\n";
my @paths = glob("$mars/resource-$res/replay-*") or ldie "cannot find any replay symlinks\n";
foreach my $path (@paths) {
my $target = readlink($path) or die "cannot read symlink '$path'\n";
my $target = readlink($path) or ldie "cannot read symlink '$path'\n";
my $nr = $target;
$nr =~ s/^log-([0-9]+)-.*$/$1/;
$min = $nr if ($nr < $min || $min < 0);
}
print "max allowed deletable logfile number: $min\n";
lprint "max allowed deletable logfile number: $min\n";
return $min;
}
@ -621,13 +655,13 @@ sub _create_delete {
$nr =~ s/^.*delete-([0-9]+)$/$1/;
}
my $new = sprintf("$mars/todo-global/delete-%09d", $nr + 1);
print "create symlink $new -> $target\n";
lprint "create symlink $new -> $target\n";
symlink($target, $new);
}
sub logdelete_res {
my ($cmd, $res) = @_;
my @paths = glob("$mars/resource-$res/log-*") or die "cannot find any logfiles\n";
my @paths = glob("$mars/resource-$res/log-*") or ldie "cannot find any logfiles\n";
@paths = sort(@paths);
my $max = _allowed_logdelete(@_);
@ -639,11 +673,11 @@ sub logdelete_res {
$nr =~ s/^.*log-([0-9]+)-.+$/$1/;
last unless $nr < $max;
print "chosen '$first' for deletion\n";
lprint "chosen '$first' for deletion\n";
_create_delete($first);
}
print "removing left-over version symlinks...\n";
lprint "removing left-over version symlinks...\n";
foreach my $versionlink (glob("$mars/resource-$res/version-*")) {
my $nrv = $versionlink;
$nrv =~ s/^.*\/version-([0-9]+)-.+$/$1/;
@ -728,36 +762,36 @@ sub fake_local_res {
my $path = "$mars/resource-$res/todo-$host/sync";
_switch($cmd, $res, $path, 0);
#check_status($res, "copy-syncstatus-$host", 0);
my $size = readlink("$mars/resource-$res/size") or die "cannot read size\n";
my $size = readlink("$mars/resource-$res/size") or ldie "cannot read size\n";
my $target = "$mars/resource-$res/syncstatus-$host";
symlink($size, "$target.tmp") or die "cannot create faked syncstatus\n";
rename("$target.tmp", $target) or die "cannot reaname symlink\n";
symlink($size, "$target.tmp") or ldie "cannot create faked syncstatus\n";
rename("$target.tmp", $target) or ldie "cannot reaname symlink\n";
}
sub _primary_res {
my ($res, $host, $pri, $old) = @_;
my $tmp = "$mars/resource-$res/.tmp.primary";
system("rm -f $tmp");
symlink($host, $tmp) or die "cannot create new primary symlink\n";
rename($tmp, $pri) or die "cannot install new primary symlink\n";
print "primary changed from '$old' to '$host'\n";
symlink($host, $tmp) or ldie "cannot create new primary symlink\n";
rename($tmp, $pri) or ldie "cannot install new primary symlink\n";
lprint "primary changed from '$old' to '$host'\n";
}
sub primary_res {
my ($cmd, $res) = @_;
my $sec = ($cmd eq "secondary");
my $pri = "$mars/resource-$res/primary";
my $old = readlink($pri) or die "cannot determine current primary\n";
my $old = readlink($pri) or ldie "cannot determine current primary\n";
if($sec) {
die "for safety reasons, switching to secondary is only allowed when I ($host) am primary\n" if($old ne $host);
ldie "for safety reasons, switching to secondary is only allowed when I ($host) am primary\n" if($old ne $host);
$host = "(none)";
} elsif($old eq $host) {
print "I am already primary.\n";
lprint "I am already primary.\n";
exit(0);
} elsif($force) {
print "FORCING myself ($host) to primary...\n";
lprint "FORCING myself ($host) to primary...\n";
} else { # try to switch myself to primary
print "trying to switch $host to primary...\n";
lprint "trying to switch $host to primary...\n";
check_sync_finished($res, $host);
check_todo($res, "connect", 1, 1);
_primary_res($res, "(none)", $pri, $old) unless $old eq "(none)";
@ -770,26 +804,26 @@ sub primary_res {
sub invalidate_res {
my ($cmd, $res) = @_;
check_not_primary(@_);
my $old_replay = readlink("$mars/resource-$res/replay-$host") or die "cannot read my own replay status symlink\n";
my $old_replay = readlink("$mars/resource-$res/replay-$host") or ldie "cannot read my own replay status symlink\n";
$old_replay =~ s/^([^,]+),.*/$1/;
my $repl = "$mars/resource-$res/todo-$host/allow-replay";
my $was_on = readlink($repl);
if ($was_on) {
_switch("pause-replay-local", $res, $repl, 0);
_trigger();
print "waiting...\n";
lprint "waiting...\n";
sleep(15);
}
my $dst = "$mars/resource-$res/syncstatus-$host";
system("rm -f $dst");
symlink("0", $dst) or die "cannot create invalidation symlink '$dst'\n";
my $primary = readlink("$mars/resource-$res/primary") or die "cannot determine primary\n";
my $replay = readlink("$mars/resource-$res/replay-$primary") or die "cannot read replay status of primary '$primary'\n";
symlink("0", $dst) or ldie "cannot create invalidation symlink '$dst'\n";
my $primary = readlink("$mars/resource-$res/primary") or ldie "cannot determine primary\n";
my $replay = readlink("$mars/resource-$res/replay-$primary") or ldie "cannot read replay status of primary '$primary'\n";
$replay =~ s/,[0-9]+,[0-9]+$/,0,0/;
_set_replaylink("$mars/resource-$res", $replay, $primary);
if ($was_on) {
_trigger();
print "waiting...\n";
lprint "waiting...\n";
sleep(15);
_switch("resume-replay-local", $res, $repl, 1);
}
@ -799,56 +833,56 @@ sub resize_res {
my ($cmd, $res) = @_;
check_primary(@_);
my @actsizes = glob("$mars/resource-$res/actsize-*");
die "resource $res has no actsize-* symlinks\n" unless @actsizes;
ldie "resource $res has no actsize-* symlinks\n" unless @actsizes;
my $lnk = "$mars/resource-$res/size";
my $old_size = readlink($lnk) or die "cannot read symlink '$lnk'\n";
my $old_size = readlink($lnk) or ldie "cannot read symlink '$lnk'\n";
my $min_size = 0;
foreach my $actsize (@actsizes) {
my $this_size = readlink($actsize) or die "cannot read symlink '$actsize'\n";
my $this_size = readlink($actsize) or ldie "cannot read symlink '$actsize'\n";
if (!$min_size || $this_size < $min_size) {
$min_size = $this_size;
}
}
print "old_size=$old_size\n";
print "new_size=$min_size\n";
die "only increases of the size are possible\n" if $min_size <= $old_size && !$force;
lprint "old_size=$old_size\n";
lprint "new_size=$min_size\n";
ldie "only increases of the size are possible\n" if $min_size <= $old_size && !$force;
foreach my $switch (glob("$mars/resource-$res/todo-*/sync")) {
my $this_switch = readlink($switch);
die "cannot read symlink '$switch'\n" unless defined($this_switch);
die "sync on '$switch' is switched on -- use marsadm pause-sync to stop\n" unless !$this_switch;
ldie "cannot read symlink '$switch'\n" unless defined($this_switch);
ldie "sync on '$switch' is switched on -- use marsadm pause-sync to stop\n" unless !$this_switch;
}
my @syncsizes = glob("$mars/resource-$res/syncstatus-$host");
foreach my $syncsize (@syncsizes) {
my $this_size = readlink($syncsize) or die "cannot read symlink '$syncsize'\n";
die "sync on $syncsize has not yet finished: $this_size != $old_size (DANGEROUS FIX: if you know what you are doing, marsadm fake-sync can 'fix' it -- but this may need a full-sync afterwards)\n" unless $this_size == $old_size;
my $this_size = readlink($syncsize) or ldie "cannot read symlink '$syncsize'\n";
ldie "sync on $syncsize has not yet finished: $this_size != $old_size (DANGEROUS FIX: if you know what you are doing, marsadm fake-sync can 'fix' it -- but this may need a full-sync afterwards)\n" unless $this_size == $old_size;
}
foreach my $syncsize (@syncsizes) {
my $this_size = readlink($syncsize) or die "cannot read symlink '$syncsize'\n";
my $this_size = readlink($syncsize) or ldie "cannot read symlink '$syncsize'\n";
unlink("$syncsize.new");
symlink($min_size, "$syncsize.new") or die "cannot create size symlink '$syncsize.new'\n";
rename("$syncsize.new", $syncsize) or die "cannot create size symlink '$syncsize'\n";;
symlink($min_size, "$syncsize.new") or ldie "cannot create size symlink '$syncsize.new'\n";
rename("$syncsize.new", $syncsize) or ldie "cannot create size symlink '$syncsize'\n";;
}
unlink("$lnk.new");
symlink($min_size, "$lnk.new") or die "cannot create size symlink '$lnk.new'\n";
rename("$lnk.new", $lnk) or die "cannot create size symlink '$lnk'\n";;
symlink($min_size, "$lnk.new") or ldie "cannot create size symlink '$lnk.new'\n";
rename("$lnk.new", $lnk) or ldie "cannot create size symlink '$lnk'\n";;
}
sub role_cmd {
my ($cmd, $res) = @_;
my $pri = "$mars/resource-$res/primary";
my $old = readlink($pri) or die "cannot determine current primary\n";
my $old = readlink($pri) or ldie "cannot determine current primary\n";
if($old eq $host) {
print "primary\n";
lprint "primary\n";
} else {
print "secondary\n";
lprint "secondary\n";
}
}
sub helplist {
my $temp;
$temp = shift;
print "ERROR: $temp" if ($temp);
print "
lprint "ERROR: $temp" if ($temp);
lprint "
marsadm <command> [<resource>] [<option>]
<command> =
@ -873,8 +907,8 @@ Advanced information are also available here: http://http://wiki.intranet.1and1.
}
sub version {
print "$0 $Id\n";
#print "my IP is $ip\n";
lprint "$0 $Id\n";
#lprint "my IP is $ip\n";
exit 0;
}
@ -949,6 +983,8 @@ foreach my $arg (@ARGV) {
my $cmd = shift @args || helplist "command argument is missing\n";
$notify = "(cmd: $cmd)" unless $cmd eq "version";
if($cmd =~ m/^help$/ || $cmd =~ m/^h$/) {
helplist;
}
@ -957,7 +993,7 @@ if($cmd =~ m/^version$/ || $cmd =~ m/^v$/) {
version;
}
die "only root may use this tool\n" if $< != 0; # getpid() seems to be missing in perlfunc
ldie "only root may use this tool\n" if $< != 0; # getpid() seems to be missing in perlfunc
helplist "unknown command $cmd\n" if !exists $cmd_table{$cmd};
my $res = "";
@ -967,7 +1003,7 @@ unless($cmd =~ m/^(create|leave)-cluster$/) {
}
print "using FORCE option -- hopefully you know what you do!\n" if $force;
lprint "using FORCE option -- hopefully you know what you do!\n" if $force;
sub do_res {
my $cmd = shift;
@ -984,7 +1020,7 @@ if($res eq "all") {
foreach $res (glob("$mars/resource-*")) {
next unless -e "$res/data-$host";
$res =~ s/^.*\/resource-(.*)$/$1/;
print "--------- resource $res\n";
lprint "--------- resource $res\n";
do_res($cmd, $res, @args);
}
} else {