File:  [LON-CAPA] / doc / build / Attic / loncapasqldatabase.html
Revision 1.6: download - view: text, annotated - select for diffs
Mon Feb 12 17:38:13 2001 UTC (23 years, 5 months ago) by harris41
Branches: MAIN
CVS tags: HEAD
complete categorized and up to date now.. including a list of things
to do -Scott

    1: <HTML>
    2: <HEAD>
    3: <TITLE>LON-CAPA SQL Database Documentation</TITLE>
    4: </HEAD>
    5: <BODY>
    6: <H1>LON-CAPA SQL Database Documentation</H1>
    7: <P>
    8: Scott Harrison
    9: </P>
   10: <P>
   11: Last updated: 02/12/2001
   12: </P>
   13: <P>
   14: This file describes issues associated with LON-CAPA
   15: and a SQL database.
   16: </P>
   17: <H2>Latest HOWTO</H2>
   18: <P>
   19: <UL>
   20: <LI>Current status of documentation</LI>
   21: <LI>Current status of implementation</LI>
   22: <LI>Purpose within LON-CAPA</LI>
   23: <LI>Installation</LI>
   24: <LI>Installation from source</LI>
   25: <LI>Configuration (automated)</LI>
   26: <LI>Manual configuration</LI>
   27: <LI>Testing</LI>
   28: <LI>Example sections of code relevant to LON-CAPA</LI>
   29: </UL>
   30: </P>
   31: <H2>Current status of documentation</H2>
   32: <P>
   33: I am going to begin documentation by inserting what notes
   34: I have into this file.  I will be subsequently rearranging
   35: them and editing them based on the tests that I conduct.
   36: I am trying to make sure that documentation, installation,
   37: and run-time issues are all consistent and correct.  The
   38: current status of everything is that it works and has
   39: been minimally tested, but things need to be cleaned up
   40: and checked again!
   41: </P>
   42: <H2>Current status of implementation</H2>
   43: <P>
   44: Need to
   45: <UL>
   46: <LI>Installation: Fix binary file listings for user permissions and ownership.
   47: <LI>Installation: Make sure sql server starts, and if database does not
   48: exist, then create. (/etc/rc.d).
   49: <LI>Processes: Make sure loncron initiates lonsql on library machines.
   50: <LI>Read in metadata from right place periodically.
   51: <LI>Implement tested perl module handler.
   52: </UL>
   53: <P>
   54: Right now, a lot of "feasibility" work has been done.
   55: Recipes for manual installation and configuration have
   56: been gathered.  Network connectivity of lond->lonsql->lond->lonc
   57: type tests have been performed.  A binary installation
   58: has been compiled in an RPM (LON-CAPA-mysql).
   59: The most lacking test in terms of feasibility has
   60: been looking at benchmarks to analyze the load at which
   61: the SQL database can efficiently allow many users to
   62: make simultaneous requests of the metadata database.
   63: </P>
   64: <P>
   65: Documentation has been pieced together over time.  But,
   66: as mentioned in the previous section, it needs an
   67: overhaul.
   68: </P>
   69: <P>
   70: The binary installation has some quirks associated with it.
   71: Some of the user permissions are wrong, although this is
   72: benign.  Also, other options of binary installation (such
   73: as using binary RPMs put together by others) were dismissed
   74: given the difficulty of getting differing combinations of
   75: these external RPMs to work together.
   76: </P>
   77: <P>
   78: Most configuration questions have been initially worked out
   79: to the point of getting this SQL software component working,
   80: however there may be more optimal approaches than currently
   81: exist.
   82: </P>
   83: <H2>Purpose within LON-CAPA</H2>
   84: <P>
   85: LON-CAPA is meant to distribute A LOT of educational content
   86: to A LOT of people.  It is ineffective to directly rely on contents
   87: within the ext2 filesystem to be speedily scanned for 
   88: on-the-fly searches of content descriptions.  (Simply put,
   89: it takes a cumbersome amount of time to open, read, analyze, and
   90: close thousands of files.)
   91: </P>
   92: <P>
   93: The solution is to hash-index various data fields that are
   94: descriptive of the educational resources on a LON-CAPA server
   95: machine.  Descriptive data fields are referred to as
   96: "metadata".  The question then arises as to how this metadata
   97: is handled in terms of the rest of the LON-CAPA network
   98: without burdening client and daemon processes.  I now
   99: answer this question in the format of Problem and Solution
  100: below.
  101: </P>
  102: <P>
  103: <PRE>
  104: PROBLEM SITUATION:
  105: 
  106:   If Server A wants data from Server B, Server A uses a lonc process to
  107:   send a database command to a Server B lond process.
  108:     lonc= loncapa client process    A-lonc= a lonc process on Server A
  109:     lond= loncapa daemon process
  110: 
  111:                  database command
  112:     A-lonc  --------TCP/IP----------------> B-lond
  113: 
  114:   The problem emerges that A-lonc and B-lond are kept waiting for the
  115:   MySQL server to "do its stuff", or in other words, perform the conceivably
  116:   sophisticated, data-intensive, time-sucking database transaction.  By tying
  117:   up a lonc and lond process, this significantly cripples the capabilities
  118:   of LON-CAPA servers. 
  119: 
  120:   While commercial databases have a variety of features that ATTEMPT to
  121:   deal with this, freeware databases are still experimenting and exploring
  122:   with different schemes with varying degrees of performance stability.
  123: 
  124: THE SOLUTION:
  125: 
  126:   A separate daemon process was created that B-lond works with to
  127:   handle database requests.  This daemon process is called "lonsql".
  128: 
  129:   So,
  130:                 database command
  131:   A-lonc  ---------TCP/IP-----------------> B-lond =====> B-lonsql
  132:          <---------------------------------/                |
  133:            "ok, I'll get back to you..."                    |
  134:                                                             |
  135:                                                             /
  136:   A-lond  <-------------------------------  B-lonc   <======
  137:            "Guess what? I have the result!"
  138: 
  139:   Of course, depending on success or failure, the messages may vary,
  140:   but the principle remains the same where a separate pool of children
  141:   processes (lonsql's) handle the MySQL database manipulations.
  142: </PRE>
  143: </P>
  144: <H2>Installation</H2>
  145: <P>
  146: Installation of the LON-CAPA SQL database normally occurs
  147: by default when using the LON-CAPA installation CD
  148: (see http://install.lon-capa.org).  It is installed
  149: as the LON-CAPA-mysql RPM.  This RPM encodes for the MySQL
  150: engine and related perl interfaces (Perl::DBI, Perl::Msql-Mysql).
  151: </P>
  152: <P>
  153: The three components of a MySQL installation for the
  154: LON-CAPA system are further described immediately below.
  155: <TABLE BORDER="0">
  156: <TR><TD COLSPAN="2"><STRONG>Perl::DBI module</STRONG>-
  157: the API "front-end"...</TD></TR>
  158: <TR><TD WIDTH="10%"></TD><TD>database interface module for organizing generic
  159: database commands which are independent of specific
  160: database implementation (such as MySQL, mSQL, Postgres, etc).
  161: </TD></TR>
  162: <TR><TD COLSPAN="2"><STRONG>Perl::MySQL module</STRONG>-
  163: the API "mid-section"...</TD></TR>
  164: <TR><TD WIDTH="10%"></TD><TD>the module to directly interface with the actual
  165: MySQL database engine</TD></TR>
  166: <TR><TD COLSPAN="2"><STRONG>MySQL database engine</STRONG>-
  167: the "back-end"...</TD></TR>
  168: <TR><TD WIDTH="10%"></TD><TD>the binary installation (compiled either
  169: from source or pre-compiled file listings) which provides the
  170: actual MySQL functionality on the system</TD></TR>
  171: </TABLE>
  172: </P>
  173: <H2>Installation from source</H2>
  174: <P>
  175: The following set of tarballs was found to work together
  176: properly on a LON-CAPA RedHat 6.2 system:
  177: <UL>
  178: <LI>DBI-1.13.tar.gz
  179: <LI>Msql-Mysql-modules-1.2209.tar.gz
  180: <LI>mysql-3.22.32.tar.gz
  181: </UL>
  182: </P>
  183: <P>
  184: Installation was simply a matter of following the instructions
  185: and typing the several "make" commands for each 
  186: </P>
  187: <H2>Configuration (automated)</H2>
  188: <P>
  189: Not yet developed.  This will be part of an interface
  190: present on LON-CAPA systems that can be launched by
  191: entering the command <TT>/usr/sbin/loncapaconfig</TT>.
  192: </P>
  193: <H2>Manual configuration</H2>
  194: <P>
  195: This is not complete.
  196: </P>
  197: <P>
  198: <STRONG>Starting the mysql daemon</STRONG>: Login on the Linux
  199: system as user 'www'.  Enter the command
  200: <TT>/usr/local/bin/safe_mysqld &</TT>
  201: </P>
  202: <P>
  203: <STRONG>Set a password for 'root'</STRONG>:
  204: <TT>/usr/local/bin/mysqladmin -u root password 'new-password'</TT>
  205: </P>
  206: <P>
  207: <STRONG>Adding a user</STRONG>:  Start the mysql daemon.  Login to the
  208: mysql system as root (<TT>mysql -u root -p mysql</TT>)
  209: and enter the right password (for instance 'newmysql').  Add the user
  210: www
  211: <PRE>
  212: INSERT INTO user (Host, User, Password)
  213: VALUES ('localhost','www',password('newmysql'));
  214: </PRE>
  215: </P>
  216: <P>
  217: <STRONG>Granting privileges to user 'www'</STRONG>:
  218: <PRE>
  219: GRANT ALL PRIVILEGES ON *.* TO www@localhost;
  220: FLUSH PRIVILEGES;
  221: </PRE>
  222: </P>
  223: <P>
  224: <STRONG>Set the SQL server to start upon system startup</STRONG>:
  225: Copy support-files/mysql.server to the right place on the system
  226: (/etc/rc.d/...).
  227: </P>
  228: <P>
  229: <STRONG>The Perl API</STRONG>
  230: <PRE>
  231:    $dbh = DBI->connect(	"DBI:mysql:loncapa",
  232: 			"www",
  233: 			"SOMEPASSWORD",
  234: 			{ RaiseError =>0,PrintError=>0});
  235: 
  236: There is an obvious need to CONNECT to the database, and in order to do
  237: this, there must be:
  238:   a RUNNING mysql daemon;
  239:   a DATABASE named "loncapa";
  240:   a USER named "www";
  241:   and an ABILITY for LON-CAPA on one machine to access
  242:        SQL database on another machine;
  243:   
  244: So, here are some notes on implementing these configurations.
  245: 
  246: ** RUNNING mysql daemon (safe_mysqld method)
  247: 
  248: The recommended way to run the MySQL daemon is as a non-root user
  249: (probably www)...
  250: 
  251: so, 1) login as user www on the linux machine
  252:     2) start the mysql daemon as /usr/local/bin/safe_mysqld &
  253: 
  254: safe_mysqld only works if the local installation of MySQL is set to the
  255: right directory permissions which I found to be:
  256: chown www:users /usr/local/var/mysql
  257: chown www:users /usr/local/lib/mysql
  258: chown -R www:users /usr/local/mysql
  259: chown www:users /usr/local/include/mysql
  260: chown www:users /usr/local/var
  261: 
  262: ** DATABASE named "loncapa"
  263: 
  264: As user www, run this command
  265:     mysql -u root -p mysql
  266: enter the password as SOMEPASSWORD
  267: 
  268: This allows you to manually enter MySQL commands.
  269: The MySQL command to generate the loncapa DATABASE is:
  270: 
  271: CREATE DATABASE 'loncapa';
  272: 
  273: ** USER named "www"
  274: 
  275: As user www, run this command
  276:     mysql -u root -p mysql
  277: enter the password as SOMEPASSWORD
  278: 
  279: To add the user www to the MySQL server, and grant all
  280: privileges on *.* to www@localhost identified by 'SOMEPASSWORD'
  281: with grant option;
  282: 
  283: INSERT INTO user (Host, User, Password)
  284: VALUES ('localhost','www',password('SOMEPASSWORD'));
  285: 
  286: GRANT ALL PRIVILEGES ON *.* TO www@localhost;
  287: 
  288: FLUSH PRIVILEGES;
  289: 
  290: ** ABILITY for LON-CAPA machines to communicate with SQL databases on
  291:    other LON-CAPA machines
  292: 
  293: An up-to-date lond and lonsql.
  294: </PRE>
  295: </P>
  296: <H2>Testing</H2>
  297: <P>
  298: <PRE>
  299: <STRONG>** TEST the database connection with my current tester.pl code
  300: which mimics what command will eventually be sent through lonc.</STRONG>
  301: 
  302: $reply=reply(
  303:     "querysend:SELECT * FROM general_information WHERE Id='AAAAA'",$lonID);
  304: </PRE>
  305: </P>
  306: <H2>Example sections of code relevant to LON-CAPA</H2>
  307: <P>
  308: Here are excerpts of code which implement the above handling:
  309: </P>
  310: <P>
  311: <PRE>
  312: <STRONG>**LONSQL
  313: A subroutine from "lonsql" which establishes a child process for handling
  314: database interactions.</STRONG>
  315: 
  316: sub make_new_child {
  317:     my $pid;
  318:     my $sigset;
  319:     
  320:     # block signal for fork
  321:     $sigset = POSIX::SigSet->new(SIGINT);
  322:     sigprocmask(SIG_BLOCK, $sigset)
  323:         or die "Can't block SIGINT for fork: $!\n";
  324:     
  325:     die "fork: $!" unless defined ($pid = fork);
  326:     
  327:     if ($pid) {
  328:         # Parent records the child's birth and returns.
  329:         sigprocmask(SIG_UNBLOCK, $sigset)
  330:             or die "Can't unblock SIGINT for fork: $!\n";
  331:         $children{$pid} = 1;
  332:         $children++;
  333:         return;
  334:     } else {
  335:         # Child can *not* return from this subroutine.
  336:         $SIG{INT} = 'DEFAULT';      # make SIGINT kill us as it did before
  337:     
  338:         # unblock signals
  339:         sigprocmask(SIG_UNBLOCK, $sigset)
  340:             or die "Can't unblock SIGINT for fork: $!\n";
  341: 	
  342: 	
  343:         #open database handle
  344: 	# making dbh global to avoid garbage collector
  345: 	unless (
  346: 		$dbh = DBI->connect("DBI:mysql:loncapa","www","SOMEPASSWORD",{ RaiseError =>0,PrintError=>0})
  347: 		) { 
  348: 	            my $st=120+int(rand(240));
  349: 		    &logthis("<font color=blue>WARNING: Couldn't connect to database  ($st secs): $@</font>");
  350: 		    print "database handle error\n";
  351: 		    sleep($st);
  352: 		    exit;
  353: 
  354: 	  };
  355: 	# make sure that a database disconnection occurs with ending kill signals
  356: 	$SIG{TERM}=$SIG{INT}=$SIG{QUIT}=$SIG{__DIE__}=\&DISCONNECT;
  357: 
  358:         # handle connections until we've reached $MAX_CLIENTS_PER_CHILD
  359:         for ($i=0; $i < $MAX_CLIENTS_PER_CHILD; $i++) {
  360:             $client = $server->accept()     or last;
  361:             
  362:             # do something with the connection
  363: 	    $run = $run+1;
  364: 	    my $userinput = <$client>;
  365: 	    chomp($userinput);
  366: 	    	    
  367: 	    my ($conserver,$querytmp)=split(/&/,$userinput);
  368: 	    my $query=unescape($querytmp);
  369: 
  370:             #send query id which is pid_unixdatetime_runningcounter
  371: 	    $queryid = $thisserver;
  372: 	    $queryid .="_".($$)."_";
  373: 	    $queryid .= time."_";
  374: 	    $queryid .= $run;
  375: 	    print $client "$queryid\n";
  376: 	    
  377:             #prepare and execute the query
  378: 	    my $sth = $dbh->prepare($query);
  379: 	    my $result;
  380: 	    unless ($sth->execute())
  381: 	    {
  382: 		&logthis("<font color=blue>WARNING: Could not retrieve from database: $@</font>");
  383: 		$result="";
  384: 	    }
  385: 	    else {
  386: 		my $r1=$sth->fetchall_arrayref;
  387: 		my @r2; map {my $a=$_; my @b=map {escape($_)} @$a; push @r2,join(",", @b)} (@$r1);
  388: 		$result=join("&",@r2) . "\n";
  389: 	    }
  390:             &reply("queryreply:$queryid:$result",$conserver);
  391: 
  392:         }
  393:     
  394:         # tidy up gracefully and finish
  395: 	
  396:         #close the database handle
  397: 	$dbh->disconnect
  398: 	   or &logthis("<font color=blue>WARNING: Couldn't disconnect from database  $DBI::errstr ($st secs): $@</font>");
  399:     
  400:         # this exit is VERY important, otherwise the child will become
  401:         # a producer of more and more children, forking yourself into
  402:         # process death.
  403:         exit;
  404:     }
  405: }
  406: </P>
  407: <P>
  408: <STRONG>** LOND enabling of MySQL requests</STRONG>
  409: <BR />
  410: This code is part of every lond child process in the
  411: way that it parses command request syntax sent to it
  412: from lonc processes.  Based on the diagram above, querysend
  413: corresponds to B-lonc sending the result of the query.
  414: queryreply corresponds to B-lond indicating that it has
  415: received the request and will start the database transaction
  416: (it returns "ok" to
  417: A-lonc ($client)).
  418: <PRE>
  419: # ------------------------------------------------------------------- querysend
  420:                    } elsif ($userinput =~ /^querysend/) {
  421:                        my ($cmd,$query)=split(/:/,$userinput);
  422: 		       $query=~s/\n*$//g;
  423:                      print $client sqlreply("$hostid{$clientip}\&$query")."\n";
  424: # ------------------------------------------------------------------ queryreply
  425:                    } elsif ($userinput =~ /^queryreply/) {
  426:                        my ($cmd,$id,$reply)=split(/:/,$userinput); 
  427: 		       my $store;
  428:                        my $execdir=$perlvar{'lonDaemons'};
  429:                        if ($store=IO::File->new(">$execdir/tmp/$id")) {
  430: 			   print $store $reply;
  431: 			   close $store;
  432: 			   print $client "ok\n";
  433: 		       }
  434: 		       else {
  435: 			   print $client "error:$!\n";
  436: 		       }
  437: 
  438: </PRE>
  439: 
  440: </P>
  441: </BODY>
  442: </HTML>

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>