
   Synchronizing Disks for HA

   What is this document?
   This is only ONE way to keeps nodes in synchronization for a high
   availability cluster.  It suits our needs and may be a good
   starting block for others so I'm sharing it.  Your mileage may vary.

   We used this on a web server to update ~1GB worth of documents in sync
   every 10 minutes.  Depending on how quickly your files change, you may
   be able to update more data, more frequently.  I would not recommend
   using this without a private fast ethernet channel, i.e. an extra NIC
   in each node connected via a crossover cable.
   
   To determine whether this method would be appropriate for you, you may
   wish to do this:
    1. Set the minimum update rate you find acceptable (say, 3 minutes).
    2. Conservatively approximate the amount of data you would expect to
       change in 3 minutes (200MB)
    3. Assuming you can count on 6MB/sec from your fast ethernet, multiply
       this by 1/3 of your update time.  Why?  You need to account for
       rsync (relatively slow executing) to determine what needs updating
       and disk write time.  In our example, this would give us 360MB
       worth of updates per cycle.

   Of course, you can always do what I did - set it up and see if it
   updates fast enough.  The scripts I've provided will notify you if it
   kicks off again before the previous one
   finished.......

   BTW, for lots of data, or rapidly changing data, you'll want a shared
   disk or use [1]drbd.  However, we find rsync is still very useful to
   keep config files, passwd files, etc. consistent and use it in
   conjuction with drbd.
   
   One last item:  rsync with the --delete option is dangerous!  Make sure
   your command options point to the proper destination!  Test this on
   non-critical data.  And then test it again.

   What you'll need:

       1.  OpenSSH
                   You can get it at:
   [2]http://www.openssh.com/portable.html
       2.  A copy of rsync
                   Found at
   [3]http://rsync.samba.org/ftp/rsync/binaries/    -OR-
   [4]http://rsync.samba.org/ftp/rsync
      3.   Cron and stuff to mirror.

       In actuality, you don't even need SSH.  You could use rsh instead,
   if your security needs permit it.

   Installing SSH

   START HERE for source distribution:
   Untar your openssh distribution.  You might want to read the INSTALL
   file or HOWTO, but most can get by with the following:
              ./configure
        make
        make install

   START HERE for rpm distribution after running "rpm -ivh
   openssh_<version#>.rpm":
   Once this is done, make sure sshd is started on bootup.  This could be
   done via init.d scripts or by placing (if ssh is installed to the
   default location) "/usr/local/sbin/sshd" in you /etc/rc.d/rc.local
   file.  Type this in now to start it.
   NOTE:      YOU SHOULD READ THE OPENSSH DOCUMENTATION.  THE FOLLOWING
   INSTRUCTIONS WILL INSTRUCT YOU
   ON ONE WAY TO SET UP AUTO-AUTHENTICATION BETWEEN YOUR CLUSTER
   NODES....BUT YOU SHOULD UNDERSTAND WHAT
   YOU ARE DOING!!!

   Make sure that /usr/local/bin is in your path and type:  "ssh-keygen
   -d"    This will create your ssh "key".   Do not enter a passphrase
   (hit return).
   In your ~/.ssh directory, there will be two files:  "id_dsa" and
   "id_dsa.pub".  "id_dsa" is your private key, "id_dsa.pub" is your
   public key (please refer to your
   ssh documentation for explanations).  Run the following commands:

   cp id_dsa.pub authorized_keys2

   chmod 400 authorized_keys2

   Now copy the entire .ssh directory (preserving permissions) to your
   home directory on the other node in the cluster.
   You should be all set.  Try it out by typing on node A:  "ssh nodeB".
   You should be logged in to nodeB without having to type anything
   (except "yes" to accept
   the host key the very first time you connect....)


   Installing Rsync

   Well, I'm not really going to tell you how to do this.  I just used the
   rpm.  If that's not possible for you, I'm sure the good folks at samba
   will have a nice README.Install for you to follow.  However, the binary
   link in the "What you'll need" section has binaries for just about all
   flavors.  Here's the link for the rpm:
   [5]http://rsync.samba.org/ftp/rsync/binaries/redhat

   Determining your Rsync command

   For our example, let's say you have a web server cluster.  As a result,
   you need the directory tree "/html" to be current on both nodes.
   Assuming node A is the master, I would use the command from node A:
    rsync --rsh=/usr/local/bin/ssh -naurvl --delete /html/ localnetB:/html

   Let's note a few things.  First, since this will be used with cron, you
   want to be sure that you use the full path for the ssh executable (and
   the rsync executable for that matter).  Also, note the "/html/"
   syntax.  This last "/" is necessary - otherwise you'll have the tree
   "/html/html" on your slave.  Lastly, I used the "-n" option.  This is
   for a dry run.  You want to do this to make sure everything is
   copied/deleted as you would expect it.  When you put this in your
   crontab file, you'll leave off the "-n" option.  Similarly, the "-v"
   verbosity option is only for this test.  In your crontab entry, it will
   be replaced with the "-q" option for quiet.  Test the command now and
   make sure it does what you want - the "--delete" option can be
   dangerous!

   Create your sync script and crontab entry

   You now want to create the script which will run rsync every X minutes
   via cron.  Our solution has the same script running on both nodes, but
   it checks whether it is running on the current primary (the one holding
   the services) or not.  If it's running on the secondary, it quits
   immediately.  I use the following perl script, called "synch_all.pl":

   #!/usr/bin/perl
   use strict;
   use diagnostics;
   use lib "/root/scripts";
   use whohostlib;
   use EnvConfig;
   use Mail::Sendmail;
   #Take care of two-way syncing case during shutdown of one server.
   Skips the first try to sync after failover.
   if ( -e "/var/lock/subsys/mirrorstop" ) {
       unlink "/var/lock/subsys/mirrorstop";
       exit;
   }
   # Only sync if serving IP
   my $dirname = "";
   my $filename = "";
   #Get server that this is running on...
   my $me = "";
   my $other = "";
   ($me,$other) = whohostlib::whohost;
   if (&whohostlib::whostatus() == 1)
   {
       #I'm serving!  You want fries with that?

       #Only sync if other node is running heartbeat:
       my $tst = `$EnvConfig::sshpath local$other
   /etc/rc.d/init.d/heartbeat status`;
       # For 0.4.9.2 and earlier use  --->    if ($tst =~ /running.../) {
       # For 0.4.9 "beta" versions...
       if ($tst =~ /heartbeat OK/) {
           #make sure previous mirror has completed...
           #To avoid hard link probs over partitions, created
   /var/lock/subsys/.DONOTREMOVE file
           if (link("/var/lock/subsys/.DONOTREMOVE",
   "/var/lock/subsys/mirror")) {
               # /home/ -  Home Directories (all users)
               # /root/scripts/ - Root scripts/system scripts - like where
   we put this script...
               # /var/spool/cron/ - System crontabs
                   my @dirlist = ( "/home/", "/root/scripts/",
   "/var/spool/cron/" );
                   foreach $dirname (@dirlist) {
                       system "/usr/bin/rsync --rsh=$EnvConfig::sshpath
   $EnvConfig::rsyncoptions $dirname local$other:$dirname";
               }
               # /etc/password - System Password File
               # /etc/shadow   - Actual Encrypted passwords.
               # /etc/group    - System Group File
               # /etc/ld.so.conf - System Linked Libraries.
               # /etc/shells     - Valid Login Shells
                   my @filelist = ( "/etc/passwd", "/etc/shadow",
   "/etc/group", "/etc/ld.so.conf", "/etc/shells" );
                   foreach $filename (@filelist) {
                       system "/usr/bin/rsync --rsh=$EnvConfig::sshpath
   $EnvConfig::rsyncoptions $filename local$other:$filename";
                   }
               # release rsync process
               unlink "/var/lock/subsys/mirror";
           } else {
               my $ddd = `date`;
               my $subject = "Next RSYNC process starting before previous
   has completed!\n";
               my $message = $subject.$ddd;
               sendmail ('smtp' => $EnvConfig::smtphost,
                   'To' => 'admin@domain.com,admin2@domain.com',
                   'From' => 'cluster@domain.com',
                   'Reply-To' => 'cluster@domain.com',
                   'Subject' => $subject,
                   'Message' => $message)
               or warn "Next RSYNC process starting before previous has
   completed (Couldn't send email)";
               #Try to get to known state by killing all synch procs and
   removing link...
               unlink "/var/lock/subsys/mirror";
               exec "/root/scripts/kill_synchs.sh";
               #"kill_synchs.sh is a one liner:  ps ax | grep synch_all |
   awk {'print $1'} | xargs -n1 kill -9
               exit;
           }
       }
   }

   Mail::Sendmail is a standard Perl module you can get from [6]CPAN, but
   you'll notice two homemade ones.  There's no magic here, just decided
   to use them for portability and readability (yes, I know, even though
   it's perl ;-)).  The "EnvConfig" one lets us set the paths and options
   we're using external to the script so that the
   actual synch_all.pl script doesn't change.  You could even add the file
   and directory lists to this, but ours doesn't change much cluster to
   cluster.  The "whohostlib" has two functions used often.  One checks
   which host the script is running on, and the other (whostatus) checks
   whether the node running the script is holding the services.

   # Configuration file for all scripts, including host definitions, etc.
   package EnvConfig;
   #www_host = "wwwdev.domain.com";
   $www_host = "wwwint.domain.com";
   #www_host = "wwwprod.domain.com";
   $sshpath = "/usr/bin/ssh";
   $rsyncoptions = "-aurlq --delete";
   #$rsyncoptions = "-aurlv";
   $smtphost = 'mail.domain.com';
   $Debug = 0;
   1;

   #!/usr/bin/perl
   use strict;
   use diagnostics;
   package whohostlib;
   sub whohost {
       my $other;
       my $me;
       #Who am I?
       $me = `hostname -s`;
       chomp $me;
       $me =~ s/[a-zA-Z]//g;

       if ($me) {
          $other = 0;
       } else {
          $other = 1;
       }
       return($me,$other);
   }
   #By doing "grep mirror haresources", we peel off the line with the info
   we want...the cluster IP.  Mirror will be added to you haresources
   later...
   sub whostatus {
       my $status = 0;
       my $x = `grep mirror /etc/ha.d/haresources`;
       my @p = split(' ',$x);
       my $ip = $p[1];
       $ip =~ s/IPaddr:://g;
       $x = `/etc/ha.d/resource.d/IPaddr $ip status`;
       # If I am primary, return 1 else return 0
       if ($x =~ /running/) {
          $status = 1;
       } else {
          $status = 0;
       }
       return $status;
   }
   1; #Return True Value

   NOTE:  We use a convention for our clusters that if the clustername is
   "foo" the two nodes that make up the cluster will be named "foo0" and
   "foo1".  The private data interfaces are given the hostnames "local0"
   and "local1".  This makes stuff easier with scripting, but you don't
   have to do it.
   NOTE #2:  You may not want a mail fired off.  You may just want a log
   entry.  You might want both.  If you're unfamiliar with perl, to add a
   log entry, substitute all the commands with "MAIL" in them above with
   the following:

   open(LOG,">>/var/log/ha-log");
   $dstr = `date +%Y/%m/%d_%T`;
   chomp $dstr;
   print LOG "$dstr RSYNC: Process starting before previous one
   completes!\n";
   close LOG;

   At this point you want to create your crontab entry.  The hard link
   prevents more than one sync process from running at the same time, but
   you want to have a decent idea of how often you need to synchronize and
   how long it will take to synchronize.

   SO, once you determine how often you'll be synchronizing, type "crontab
   -e" to modify your crontab entry.  If you don't like vi, try using
   "setenv EDITOR /usr/local/bin/emacs" (or export for bash users) to
   select emacs or a different editor.  If you want to synchronize every
   10 minutes, your entry would look like this:

*/10 * * * * /script_directory/sync.pl &> /dev/null

   You could also redirect output to some logfile if you desire, but keep
   in mind how often it runs.

   Dealing with Failover

   We're just about there now.  The last thing you need to consider is
   when you shutdown.  You want to do one last synchronization before you
   shut down.  I accomplished this with a heartbeat service called
   "mirror".  On startup, we do nothing other than prevent the very first
   synch_all (as a precaution) and notify the necessary admins about the
   failover.

   NOTE:   You will want to be sure that any applications writing to the
   synch'ed areas stop before your last sync.  If these applications are
   controlled via the ipresources configuration file, you can ensure this
   by listing the "mirror" script right after your IP address.  However,
   if they are not, you may want to add an application exit to the
   beginning of your mirror script.

   #!/usr/bin/perl
   use lib "/root/scripts";
   use whohostlib;
   # See how we were called.
   if ($#ARGV == 0) {
       $switch = $ARGV[0];
   } else {
       print "Usage: mirror {start|stop|status|restart}\n";
       exit -1;
   }
   #  start)
   if ($switch eq "start") {
       #Make sure to skip first sync.....
       ($me,$other) = whohostlib::whohost;
       system ("/usr/bin/ssh local$me /bin/touch
   /var/lock/subsys/mirrorstop");
       #Notify admins of startup....
       open (XXX,"/root/scripts/.pagelist");
       $hh = `hostname`;
       chomp $hh;
       while (<XXX>) {
       chomp;
       system "/root/scripts/pagescript.pl $_ \"Starting up heartbeat
   services on node $hh\"";
       }
       close XXX;

   } elsif ($switch eq "stop") {
       #stop)
       print "Mirror stop: \n";
       #Am I serving???
       #We'll consider mirror to be "running" if the IP is on this box...
       $x = `grep mirror /etc/ha.d/haresources`;
       @p = split(' ',$x);
       $ip = $p[1];
       $x = `/etc/ha.d/resource.d/IPaddr $ip status`;
       if ($x =~ /running/) {
       #Notify admins of shutdown....
       open (XXX,"/root/scripts/.pagelist");
       $hh = `hostname`;
       chomp $hh;
       while (<XXX>) {
           chomp;
           system "/root/scripts/pagescript.pl $_ \"Shutting down
   heartbeat services on node $hh\"";
       }
       close XXX;
       #You want fries with that?  One last sync to other box, if not in
   the middle of one now...
       if ( -e "/var/lock/subsys/mirror" ) {
           print "Not syncing, already doing so...\n";
       } else {
           print "Synchronizing data on standby node: \n";
           system "/root/scripts/synch_all.pl";
       }
       # To prevent NEW master from syncing back to us too soon, lock
   their first synch_all...
       ($me,$other) = whohostlib::whohost;
       system ("/usr/bin/ssh local$other /bin/touch
   /var/lock/subsys/mirrorstop");
       }

   } elsif ($switch eq "status") {
       #status)
       #We'll consider mirror to be "running" if the IP is on this box...
       $y = `grep mirror /etc/ha.d/haresources`;
       @p = split(' ',$y);
       $ip = $p[1];

       $x = `/etc/ha.d/resource.d/IPaddr $ip status`;
       print $x;
   } elsif ($switch eq "restart") {
       #restart)

       system "/etc/ha.d/resource.d/mirror stop";
       system "/etc/ha.d/resource.d/mirror start";

   } else {
       print "Usage: mirror {start|stop|status|restart}\n";
   }
   exit;

   Finally, you need to install the script in your ipresources on both
   nodes.  You'll want mirror to be the first service listed (after IP).
   For our webserver, it would read:
    nodeA 192.168.85.1 mirror httpd



   Be Careful

   Please test your setup on non-critical data first.  There could be a
   bad typo above or whatever.  The "--delete" option can be dangerous.
   You've been warned.

   Rev 1.0.0
   Rudy Pawul
   rpawul@iso-ne.com

References

   1. http://www.complang.tuwien.ac.at/reisner/drbd/
   2. http://www.openssh.com/portable.html
   3. http://rsync.samba.org/ftp/rsync/binaries/
   4. http://rsync.samba.org/ftp/rsync/rsync-2.3.1.tar.gz
   5. http://rsync.samba.org/ftp/rsync/binaries/redhat
   6. file://localhost/usr/obj/ports/heartbeat-2.1.2-15/Heartbeat-Dev-c492f19cb583/doc/www.cpan.org
