2 #============================================================= -*-perl-*-
4 # BackupPC: Main program for PC backups.
8 # BackupPC reads the configuration and status information from
9 # $ConfDir/conf. It then runs and manages all the backup activity.
11 # As specified by $Conf{WakeupSchedule}, BackupPC wakes up periodically
12 # to queue backups on all the PCs. This is a four step process:
13 # 1) For each host and DHCP address backup requests are queued on the
14 # background command queue.
15 # 2) For each PC, BackupPC_dump is forked. Several of these may
16 # be run in parallel, based on the configuration.
17 # 3) For each complete, good, backup, BackupPC_link is forked.
18 # Only one of these tasks runs at a time.
19 # 4) In the background BackupPC_trashClean is run to remove any expired
20 # backups. Once each night, BackupPC_nightly is run to complete some
21 # additional administrative tasks (pool cleaning etc).
23 # BackupPC also listens for connections on a unix domain socket and
24 # the tcp port $Conf{ServerPort}, which are used by various
25 # sub-programs and the CGI script BackupPC_Admin for status reporting
26 # and user-initiated backup or backup cancel requests.
29 # Craig Barratt <cbarratt@users.sourceforge.net>
32 # Copyright (C) 2001-2007 Craig Barratt
34 # This program is free software; you can redistribute it and/or modify
35 # it under the terms of the GNU General Public License as published by
36 # the Free Software Foundation; either version 2 of the License, or
37 # (at your option) any later version.
39 # This program is distributed in the hope that it will be useful,
40 # but WITHOUT ANY WARRANTY; without even the implied warranty of
41 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
42 # GNU General Public License for more details.
44 # You should have received a copy of the GNU General Public License
45 # along with this program; if not, write to the Free Software
46 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
48 #========================================================================
50 # Version 3.2.0beta0, released 17 Jan 2009.
52 # See http://backuppc.sourceforge.net.
54 #========================================================================
58 use vars qw(%Status %Info $Hosts);
59 use lib "/usr/local/BackupPC/lib";
61 use BackupPC::FileZIO;
62 use Encode qw/decode_utf8/;
72 ###########################################################################
73 # Handle command line options
74 ###########################################################################
76 if ( !getopts("d", \%opts) || @ARGV != 0 ) {
77 print("usage: $0 [-d]\n");
81 ###########################################################################
82 # Initialize major data structures and variables
83 ###########################################################################
86 # Get an instance of BackupPC::Lib and get some shortcuts.
88 die("BackupPC::Lib->new failed\n") if ( !(my $bpc = BackupPC::Lib->new) );
89 my $TopDir = $bpc->TopDir();
90 my $BinDir = $bpc->BinDir();
91 my $LogDir = $bpc->LogDir();
92 my %Conf = $bpc->Conf();
95 # Verify we are running as the correct user
97 if ( $Conf{BackupPCUserVerify}
98 && $> != (my $uid = (getpwnam($Conf{BackupPCUser}))[2]) ) {
99 die "Wrong user: my userid is $>, instead of $uid ($Conf{BackupPCUser})\n";
103 # %Status maintain status information about each host.
104 # It is a hash of hashes, whose first index is the host.
109 # %Info is a hash giving general information about BackupPC status.
116 if ( -f "$LogDir/status.pl" && !(my $ret = do "$LogDir/status.pl") ) {
118 print STDERR "couldn't parse $LogDir/status.pl: $@";
119 } elsif ( !defined($ret) ) {
120 print STDERR "couldn't do $LogDir/status.pl: $!";
122 print STDERR "couldn't run $LogDir/status.pl";
127 # %Jobs maintains information about currently running jobs.
128 # It is a hash of hashes, whose first index is the host.
133 # There are three command queues:
134 # - @UserQueue is a queue of user initiated backup requests.
135 # - @BgQueue is a queue of automatically scheduled backup requests.
136 # - @CmdQueue is a queue of administrative jobs, including tasks
137 # like BackupPC_link, BackupPC_trashClean, and BackupPC_nightly
138 # Each queue is an array of hashes. Each hash stores information
139 # about the command request.
146 # To quickly lookup if a given host is on a given queue, we keep
147 # a hash of flags for each queue type.
149 my(%CmdQueueOn, %UserQueueOn, %BgQueueOn);
152 # One or more clients can connect to the server to get status information
153 # or request/cancel backups etc. The %Clients hash maintains information
154 # about each of these socket connections. The hash key is an incrementing
155 # number stored in $ClientConnCnt. Each entry is a hash that contains
156 # various information about the client connection.
162 # Read file descriptor mask used by select(). Every file descriptor
163 # on which we expect to read (or accept) has the corresponding bit
169 # Unix seconds when we next wakeup. A value of zero forces the scheduler
170 # to compute the next wakeup time.
175 # Name of signal saved by catch_signal
182 my($RunNightlyWhenIdle, $FirstWakeup, $CmdJob, $ServerInetPort);
183 my($BackupPCNightlyJobs, $BackupPCNightlyLock);
186 # Complete the rest of the initialization
190 ###########################################################################
192 ###########################################################################
196 # Check if we can/should run BackupPC_nightly
198 Main_TryToRun_nightly();
201 # Check if we can run a new command from @CmdQueue.
203 Main_TryToRun_CmdQueue();
206 # Check if we can run a new command from @UserQueue or @BgQueue.
208 Main_TryToRun_Bg_or_User_Queue();
211 # Do a select() to wait for the next interesting thing to happen
212 # (timeout, signal, someone sends a message, child dies etc).
214 my $fdRead = Main_Select();
217 # Process a signal if we received one.
220 Main_Process_Signal();
225 # Check if a timeout has occurred.
227 Main_Check_Timeout();
230 # Check for, and process, any messages (output) from our jobs
232 Main_Check_Job_Messages($fdRead);
235 # Check for, and process, any output from our clients. Also checks
236 # for new connections to our SERVER_UNIX and SERVER_INET sockets.
238 Main_Check_Client_Messages($fdRead);
241 ############################################################################
244 # Main initialization routine. Called once at statup.
245 ############################################################################
248 umask($Conf{UmaskMode});
251 # Check for another running process, verify executables are configured
252 # correctly and make sure $TopDir is on a file system that supports
255 if ( $Info{pid} ne "" && kill(0, $Info{pid}) ) {
256 print(STDERR $bpc->timeStamp,
257 "Another BackupPC is running (pid $Info{pid}); quitting...\n");
261 foreach my $progName ( qw(SmbClientPath NmbLookupPath PingPath DfPath
262 SendmailPath SshPath) ) {
263 next if ( $Conf{$progName} eq "" || -x $Conf{$progName} );
264 print(STDERR $bpc->timeStamp,
265 "\$Conf{$progName} = '$Conf{$progName}' is not a"
266 . " valid executable program\n");
270 if ( !$bpc->HardlinkTest("$TopDir/pc", "$TopDir/cpool") ) {
271 print(STDERR $bpc->timeStamp, "Can't create a test hardlink between a file"
272 . " in $TopDir/pc and $TopDir/cpool. Either these are different"
273 . " file systems, or this file system doesn't support hardlinks,"
274 . " or these directories don't exist, or there is a permissions"
275 . " problem, or the file system is out of inodes or full. Use"
276 . " df, df -i, and ls -ld to check each of these possibilities."
283 # daemonize by forking; more robust method per:
284 # http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=301057
287 defined($pid = fork) or die("Can't fork: $!");
288 exit if( $pid ); # parent exits
291 defined($pid = fork) or die("Can't fork: $!");
292 exit if $pid; # parent exits
294 chdir ("/") or die("Cannot chdir to /: $!\n");
296 open(STDIN , ">/dev/null") or die("Cannot open /dev/null as stdin\n");
297 # STDOUT and STDERR are handled in LogFileOpen() right below,
298 # otherwise we would have to reopen them too.
302 # Open the LOG file and redirect STDOUT, STDERR etc
307 # Read the hosts file (force a read).
309 exit(1) if ( !HostsUpdate(1) );
312 # Clean up %ENV for taint checking
314 delete @ENV{qw(IFS CDPATH ENV BASH_ENV)};
315 $ENV{PATH} = $Conf{MyPath};
318 # Initialize server sockets
323 # Catch various signals
325 foreach my $sig ( qw(INT BUS SEGV PIPE TERM ALRM HUP) ) {
326 $SIG{$sig} = \&catch_signal;
330 # Report that we started, and update %Info.
332 print(LOG $bpc->timeStamp, "BackupPC started, pid $$\n");
333 $Info{ConfigModTime} = $bpc->ConfigMTime();
335 $Info{startTime} = time;
336 $Info{ConfigLTime} = time;
337 $Info{Version} = $bpc->{Version};
340 # Update the status left over form the last time BackupPC ran.
341 # Requeue any pending links.
343 foreach my $host ( sort(keys(%$Hosts)) ) {
344 if ( $Status{$host}{state} eq "Status_backup_in_progress" ) {
346 # should we restart it? skip it for now.
348 $Status{$host}{state} = "Status_idle";
349 } elsif ( $Status{$host}{state} eq "Status_link_pending"
350 || $Status{$host}{state} eq "Status_link_running" ) {
353 $Status{$host}{state} = "Status_idle";
355 $Status{$host}{activeJob} = 0;
357 foreach my $host ( sort(keys(%Status)) ) {
358 next if ( defined($Hosts->{$host}) );
359 delete($Status{$host});
363 # Write out our initial status and save our PID
366 unlink("$LogDir/BackupPC.pid");
367 if ( open(PID, ">", "$LogDir/BackupPC.pid") ) {
370 chmod(0444, "$LogDir/BackupPC.pid");
374 # For unknown reasons there is a very infrequent error about not
375 # being able to coerce GLOBs inside the XS Data::Dumper. I've
376 # only seen this on a particular platform and perl version.
377 # For now the workaround appears to be use the perl version of
380 $Data::Dumper::Useqq = 1;
383 ############################################################################
384 # Main_TryToRun_nightly()
386 # Checks to see if we can/should run BackupPC_nightly or
387 # BackupPC_trashClean. If so we push the appropriate command onto
389 ############################################################################
390 sub Main_TryToRun_nightly
393 # Check if we should run BackupPC_nightly or BackupPC_trashClean.
394 # BackupPC_nightly is run when the current job queue is empty.
395 # BackupPC_trashClean is run in the background always.
397 my $trashCleanRunning = defined($Jobs{$bpc->trashJob}) ? 1 : 0;
398 if ( !$trashCleanRunning && !$CmdQueueOn{$bpc->trashJob} ) {
400 # This should only happen once at startup, but just in case this
401 # code will re-start BackupPC_trashClean if it quits
404 host => $bpc->trashJob,
407 cmd => ["$BinDir/BackupPC_trashClean"],
409 $CmdQueueOn{$bpc->trashJob} = 1;
411 if ( $RunNightlyWhenIdle == 1 ) {
413 # Queue multiple nightly jobs based on the configuration
415 $Conf{MaxBackupPCNightlyJobs} = 1
416 if ( $Conf{MaxBackupPCNightlyJobs} <= 0 );
417 $Conf{BackupPCNightlyPeriod} = 1
418 if ( $Conf{BackupPCNightlyPeriod} <= 0 );
420 # Decide what subset of the 16 top-level directories 0..9a..f
421 # we run BackupPC_nightly on, based on $Conf{BackupPCNightlyPeriod}.
422 # If $Conf{BackupPCNightlyPeriod} == 1 then we run 0..15 every
423 # time. If $Conf{BackupPCNightlyPeriod} == 2 then we run
424 # 0..7 one night and 89a-f the next night. And so on.
426 # $Info{NightlyPhase} counts which night, from 0 to
427 # $Conf{BackupPCNightlyPeriod} - 1.
429 my $start = int($Info{NightlyPhase} * 16
430 / $Conf{BackupPCNightlyPeriod});
431 my $end = int(($Info{NightlyPhase} + 1) * 16
432 / $Conf{BackupPCNightlyPeriod});
433 $end = $start + 1 if ( $end <= $start );
434 $Info{NightlyPhase}++;
435 $Info{NightlyPhase} = 0 if ( $end >= 16 );
438 # Zero out the data we expect to get from BackupPC_nightly.
440 for my $p ( qw(pool cpool) ) {
441 for ( my $i = $start ; $i < $end ; $i++ ) {
442 $Info{pool}{$p}[$i]{FileCnt} = 0;
443 $Info{pool}{$p}[$i]{DirCnt} = 0;
444 $Info{pool}{$p}[$i]{Kb} = 0;
445 $Info{pool}{$p}[$i]{Kb2} = 0;
446 $Info{pool}{$p}[$i]{KbRm} = 0;
447 $Info{pool}{$p}[$i]{FileCntRm} = 0;
448 $Info{pool}{$p}[$i]{FileCntRep} = 0;
449 $Info{pool}{$p}[$i]{FileRepMax} = 0;
450 $Info{pool}{$p}[$i]{FileCntRename} = 0;
451 $Info{pool}{$p}[$i]{FileLinkMax} = 0;
452 $Info{pool}{$p}[$i]{Time} = 0;
455 print(LOG $bpc->timeStamp,
456 sprintf("Running %d BackupPC_nightly jobs from %d..%d"
457 . " (out of 0..15)\n",
458 $Conf{MaxBackupPCNightlyJobs}, $start, $end - 1));
461 # Now queue the $Conf{MaxBackupPCNightlyJobs} jobs.
462 # The granularity on start and end is now 0..255.
467 for ( my $i = 0 ; $i < $Conf{MaxBackupPCNightlyJobs} ; $i++ ) {
469 # The first nightly job gets the -m option (does email, log aging).
470 # All jobs get the start and end options from 0..255 telling
471 # them which parts of the pool to traverse.
473 my $cmd = ["$BinDir/BackupPC_nightly"];
474 push(@$cmd, "-m") if ( $i == 0 );
476 $start = $start0 + int(($end - $start0)
477 * ($i + 1) / $Conf{MaxBackupPCNightlyJobs});
478 push(@$cmd, $start - 1);
479 my $job = $bpc->adminJob($i);
486 $CmdQueueOn{$job} = 1;
488 $RunNightlyWhenIdle = 2;
492 ############################################################################
493 # Main_TryToRun_CmdQueue()
495 # Decide if we can run a new command from the @CmdQueue.
496 # We only run one of these at a time. The @CmdQueue is
497 # used to run BackupPC_link (for the corresponding host),
498 # BackupPC_trashClean, and BackupPC_nightly using a fake
499 # host name of $bpc->adminJob.
500 ############################################################################
501 sub Main_TryToRun_CmdQueue
505 while ( $CmdJob eq "" && @CmdQueue > 0 && $RunNightlyWhenIdle != 1
506 || @CmdQueue > 0 && $RunNightlyWhenIdle == 2
507 && $bpc->isAdminJob($CmdQueue[0]->{host})
510 $req = pop(@CmdQueue);
512 $host = $req->{host};
513 if ( defined($Jobs{$host}) ) {
514 print(LOG $bpc->timeStamp,
515 "Botch on admin job for $host: already in use!!\n");
517 # This could happen during normal opertion: a user could
518 # request a backup while a BackupPC_link is queued from
519 # a previous backup. But it is unlikely. Just put this
520 # request back on the end of the queue.
522 unshift(@CmdQueue, $req);
525 $CmdQueueOn{$host} = 0;
526 my $cmd = $req->{cmd};
527 my $pid = open(FH, "-|");
528 if ( !defined($pid) ) {
529 print(LOG $bpc->timeStamp,
530 "can't fork for $host, request by $req->{user}\n");
536 $ENV{BPC_REQUSER} = $req->{user};
537 POSIX::nice($Conf{CmdQueueNice}) if ( $Conf{CmdQueueNice} );
539 print(LOG $bpc->timeStamp, "can't exec @$cmd for $host\n");
542 $Jobs{$host}{pid} = $pid;
543 $Jobs{$host}{fh} = *FH;
544 $Jobs{$host}{fn} = fileno(FH);
545 vec($FDread, $Jobs{$host}{fn}, 1) = 1;
546 $Jobs{$host}{startTime} = time;
547 $Jobs{$host}{reqTime} = $req->{reqTime};
548 $cmd = $bpc->execCmd2ShellCmd(@$cmd);
549 $Jobs{$host}{cmd} = $cmd;
550 $Jobs{$host}{user} = $req->{user};
551 $Jobs{$host}{type} = $Status{$host}{type};
552 $Status{$host}{state} = "Status_link_running";
553 $Status{$host}{activeJob} = 1;
554 $Status{$host}{endTime} = time;
555 $CmdJob = $host if ( $host ne $bpc->trashJob );
556 $cmd =~ s/$BinDir\///g;
557 print(LOG $bpc->timeStamp, "Running $cmd (pid=$pid)\n");
558 if ( $cmd =~ /^BackupPC_nightly\s/ ) {
559 $BackupPCNightlyJobs++;
560 $BackupPCNightlyLock++;
565 ############################################################################
566 # Main_TryToRun_Bg_or_User_Queue()
568 # Decide if we can run any new backup requests from @BgQueue
569 # or @UserQueue. Several of these can be run at the same time
570 # based on %Conf settings. Jobs from @UserQueue take priority,
571 # and at total of $Conf{MaxBackups} + $Conf{MaxUserBackups}
572 # simultaneous jobs can run from @UserQueue. After @UserQueue
573 # is exhausted, up to $Conf{MaxBackups} simultaneous jobs can
575 ############################################################################
576 sub Main_TryToRun_Bg_or_User_Queue
579 my(@deferUserQueue, @deferBgQueue);
582 if ( time - $Info{DUlastValueTime} >= 600 ) {
584 # Update our notion of disk usage no more than
585 # once every 10 minutes
587 $du = $bpc->CheckFileSystemUsage($TopDir);
588 $Info{DUlastValue} = $du;
589 $Info{DUlastValueTime} = time;
592 # if we recently checked it then just use the old value
594 $du = $Info{DUlastValue};
596 if ( $Info{DUDailyMaxReset} ) {
597 $Info{DUDailyMaxStartTime} = time;
598 $Info{DUDailyMaxReset} = 0;
599 $Info{DUDailyMax} = 0;
601 if ( $du > $Info{DUDailyMax} ) {
602 $Info{DUDailyMax} = $du;
603 $Info{DUDailyMaxTime} = time;
605 if ( $du > $Conf{DfMaxUsagePct} ) {
606 my @bgQueue = @BgQueue;
610 # When the disk is too full, only run backups that will
611 # do expires, not regular backups
614 foreach $req ( @bgQueue ) {
615 if ( $req->{dumpExpire} ) {
616 unshift(@BgQueue, $req);
618 $BgQueueOn{$req->{host}} = 0;
623 print(LOG $bpc->timeStamp,
624 "Disk too full ($du%); skipped $nSkip hosts\n");
625 $Info{DUDailySkipHostCnt} += $nSkip;
630 # Run background jobs anytime. Previously they were locked out
631 # when BackupPC_nightly was running or pending with this
632 # condition on the while loop:
634 # while ( $RunNightlyWhenIdle == 0 )
638 my(@args, $progName, $type);
639 my $nJobs = keys(%Jobs);
641 # CmdJob and trashClean don't count towards MaxBackups / MaxUserBackups
643 if ( $CmdJob ne "" ) {
644 if ( $BackupPCNightlyJobs ) {
645 $nJobs -= $BackupPCNightlyJobs;
650 $nJobs-- if ( defined($Jobs{$bpc->trashJob} ) );
651 if ( $nJobs < $Conf{MaxBackups} + $Conf{MaxUserBackups}
652 && @UserQueue > 0 ) {
653 $req = pop(@UserQueue);
654 if ( defined($Jobs{$req->{host}}) ) {
655 push(@deferUserQueue, $req);
658 push(@args, $req->{doFull} ? "-f" : "-i")
659 if (( !$req->{restore} ) && ( !$req->{archive} ));
660 $UserQueueOn{$req->{host}} = 0;
661 } elsif ( $nJobs < $Conf{MaxBackups}
662 && (@CmdQueue + $nJobs)
663 <= $Conf{MaxBackups} + $Conf{MaxPendingCmds}
665 $req = pop(@BgQueue);
666 if ( defined($Jobs{$req->{host}}) ) {
668 # Job is currently running for this host; save it for later
670 unshift(@deferBgQueue, $req);
673 $BgQueueOn{$req->{host}} = 0;
676 # Restore the deferred jobs
678 @BgQueue = (@BgQueue, @deferBgQueue);
679 @UserQueue = (@UserQueue, @deferUserQueue);
682 $host = $req->{host};
683 my $user = $req->{user};
684 if ( $req->{restore} ) {
685 $progName = "BackupPC_restore";
687 push(@args, $req->{hostIP}, $req->{host}, $req->{reqFileName});
688 } elsif ( $req->{archive} ) {
689 $progName = "BackupPC_archive";
691 push(@args, $req->{user}, $req->{host}, $req->{reqFileName});
693 $progName = "BackupPC_dump";
695 push(@args, "-d") if ( $req->{dhcp} );
696 push(@args, "-e") if ( $req->{dumpExpire} );
699 my $pid = open(FH, "-|");
700 if ( !defined($pid) ) {
701 print(LOG $bpc->timeStamp,
702 "can't fork to run $progName for $host, request by $user\n");
708 exec("$BinDir/$progName", @args);
709 print(LOG $bpc->timeStamp, "can't exec $progName for $host\n");
712 $Jobs{$host}{pid} = $pid;
713 $Jobs{$host}{fh} = *FH;
714 $Jobs{$host}{fn} = fileno(FH);
715 $Jobs{$host}{dhcp} = $req->{dhcp};
716 vec($FDread, $Jobs{$host}{fn}, 1) = 1;
717 $Jobs{$host}{startTime} = time;
718 $Jobs{$host}{reqTime} = $req->{reqTime};
719 $Jobs{$host}{userReq} = $req->{userReq};
720 $Jobs{$host}{cmd} = $bpc->execCmd2ShellCmd($progName, @args);
721 $Jobs{$host}{user} = $user;
722 $Jobs{$host}{type} = $type;
723 $Status{$host}{userReq} = $req->{userReq}
724 if ( defined($Hosts->{$host}) );
725 if ( !$req->{dhcp} ) {
726 $Status{$host}{state} = "Status_".$type."_starting";
727 $Status{$host}{activeJob} = 1;
728 $Status{$host}{startTime} = time;
729 $Status{$host}{endTime} = "";
734 ############################################################################
737 # If necessary, figure out when to next wakeup based on $Conf{WakeupSchedule},
738 # and then do a select() to wait for the next thing to happen
739 # (timeout, signal, someone sends a message, child dies etc).
740 ############################################################################
743 if ( $NextWakeup <= 0 ) {
745 # Figure out when to next wakeup based on $Conf{WakeupSchedule}.
747 my($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst)
749 my($currHours) = $hour + $min / 60 + $sec / 3600;
750 if ( $bpc->ConfigMTime() != $Info{ConfigModTime} ) {
751 ServerReload("Re-read config file because mtime changed");
754 foreach my $t ( @{$Conf{WakeupSchedule} || [0..23]} ) {
755 next if ( $t < 0 || $t > 24 );
756 my $tomorrow = $t + 24;
758 || ($tomorrow - $currHours > 0
759 && $delta > $tomorrow - $currHours) ) {
760 $delta = $tomorrow - $currHours;
761 $FirstWakeup = $t == $Conf{WakeupSchedule}[0];
764 || ($t - $currHours > 0 && $delta > $t - $currHours) ) {
765 $delta = $t - $currHours;
766 $FirstWakeup = $t == $Conf{WakeupSchedule}[0];
769 $NextWakeup = time + $delta * 3600;
770 $Info{nextWakeup} = $NextWakeup;
771 print(LOG $bpc->timeStamp, "Next wakeup is ",
772 $bpc->timeStamp($NextWakeup, 1), "\n");
775 # Call select(), waiting until either a signal, a timeout,
776 # any output from our jobs, or any messages from clients
778 # select() is where we (hopefully) spend most of our time blocked...
780 my $timeout = $NextWakeup - time;
781 $timeout = 1 if ( $timeout <= 0 );
783 select(my $rout = $FDread, undef, $ein, $timeout);
788 ############################################################################
789 # Main_Process_Signal()
792 ############################################################################
793 sub Main_Process_Signal
798 if ( $SigName eq "HUP" ) {
799 ServerReload("Re-read config file because of a SIG_HUP");
800 } elsif ( $SigName ) {
801 ServerShutdown("Got signal $SigName... cleaning up");
806 ############################################################################
807 # Main_Check_Timeout()
809 # Check if a timeout has occured, and if so, queue all the PCs for backups.
810 # Also does log file aging on the first timeout after midnight.
811 ############################################################################
812 sub Main_Check_Timeout
817 return if ( time < $NextWakeup || $NextWakeup <= 0 );
819 if ( $FirstWakeup ) {
821 # This is the first wakeup after midnight. Do log file aging
822 # and various house keeping.
825 printf(LOG "%s24hr disk usage: %d%% max, %d%% recent,"
826 . " %d skipped hosts\n",
827 $bpc->timeStamp, $Info{DUDailyMax}, $Info{DUlastValue},
828 $Info{DUDailySkipHostCnt});
829 $Info{DUDailyMaxReset} = 1;
830 $Info{DUDailyMaxPrev} = $Info{DUDailyMax};
831 $Info{DUDailySkipHostCntPrev} = $Info{DUDailySkipHostCnt};
832 $Info{DUDailySkipHostCnt} = 0;
833 my $lastLog = $Conf{MaxOldLogFiles} - 1;
834 if ( -f "$LogDir/LOG.$lastLog" ) {
835 print(LOG $bpc->timeStamp,
836 "Removing $LogDir/LOG.$lastLog\n");
837 unlink("$LogDir/LOG.$lastLog");
839 if ( -f "$LogDir/LOG.$lastLog.z" ) {
840 print(LOG $bpc->timeStamp,
841 "Removing $LogDir/LOG.$lastLog.z\n");
842 unlink("$LogDir/LOG.$lastLog.z");
844 print(LOG $bpc->timeStamp, "Aging LOG files, LOG -> LOG.0 -> "
845 . "LOG.1 -> ... -> LOG.$lastLog\n");
846 close(STDERR); # dup of LOG
847 close(STDOUT); # dup of LOG
849 for ( my $i = $lastLog - 1 ; $i >= 0 ; $i-- ) {
851 rename("$LogDir/LOG.$i", "$LogDir/LOG.$j")
852 if ( -f "$LogDir/LOG.$i" );
853 rename("$LogDir/LOG.$i.z", "$LogDir/LOG.$j.z")
854 if ( -f "$LogDir/LOG.$i.z" );
857 # Compress the log file LOG -> LOG.0.z (if enabled).
858 # Otherwise, just rename LOG -> LOG.0.
860 BackupPC::FileZIO->compressCopy("$LogDir/LOG",
863 $Conf{CompressLevel}, 1);
866 # Remember to run the nightly script when the next CmdQueue
869 if ( $RunNightlyWhenIdle == 2 ) {
870 print(LOG $bpc->timeStamp, "BackupPC_nightly is still running after 24 hours!!"
871 . " You should adjust the config settings; Skipping this run\n");
873 $RunNightlyWhenIdle = 1;
877 # Write out the current status and then queue all the PCs
881 %BgQueueOn = () if ( @BgQueue == 0 );
882 %UserQueueOn = () if ( @UserQueue == 0 );
883 %CmdQueueOn = () if ( @CmdQueue == 0 );
887 ############################################################################
888 # Main_Check_Job_Messages($fdRead)
890 # Check if select() says we have bytes waiting from any of our jobs.
891 # Handle each of the messages when complete (newline terminated).
892 ############################################################################
893 sub Main_Check_Job_Messages
896 foreach my $host ( keys(%Jobs) ) {
897 next if ( !vec($fdRead, $Jobs{$host}{fn}, 1) );
900 # do a last check to make sure there is something to read so
901 # we are absolutely sure we won't block.
903 vec(my $readMask, $Jobs{$host}{fn}, 1) = 1;
904 if ( !select($readMask, undef, undef, 0.0) ) {
905 print(LOG $bpc->timeStamp, "Botch in Main_Check_Job_Messages:"
906 . " nothing to read from $host. Debug dump:\n");
907 my($dump) = Data::Dumper->new(
908 [ \%Clients, \%Jobs, \$FDread, \$fdRead],
909 [qw(*Clients, *Jobs *FDread, *fdRead)]);
911 print(LOG $dump->Dump);
914 my $nbytes = sysread($Jobs{$host}{fh}, $mesg, 1024);
915 $Jobs{$host}{mesg} .= $mesg if ( $nbytes > 0 );
917 # Process any complete lines of output from this jobs.
918 # Any output to STDOUT or STDERR from the children is processed here.
920 while ( $Jobs{$host}{mesg} =~ /(.*?)[\n\r]+(.*)/s ) {
922 $Jobs{$host}{mesg} = $2;
923 if ( $Jobs{$host}{dhcp} ) {
924 if ( $mesg =~ /^DHCP (\S+) (\S+)/ ) {
925 my $newHost = $bpc->uriUnesc($2);
926 if ( defined($Jobs{$newHost}) ) {
927 print(LOG $bpc->timeStamp,
928 "Backup on $newHost is already running\n");
929 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
933 $Jobs{$host}{dhcpHostIP} = $host;
934 $Status{$newHost}{dhcpHostIP} = $host;
935 $Jobs{$newHost} = $Jobs{$host};
936 delete($Jobs{$host});
938 $Status{$host}{state} = "Status_backup_starting";
939 $Status{$host}{activeJob} = 1;
940 $Status{$host}{startTime} = $Jobs{$host}{startTime};
941 $Status{$host}{endTime} = "";
942 $Jobs{$host}{dhcp} = 0;
944 print(LOG $bpc->timeStamp, "dhcp $host: $mesg\n");
946 } elsif ( $mesg =~ /^started (.*) dump, share=(.*)/ ) {
947 $Jobs{$host}{type} = $1;
948 $Jobs{$host}{shareName} = $2;
949 print(LOG $bpc->timeStamp,
950 "Started $1 backup on $host (pid=$Jobs{$host}{pid}",
951 $Jobs{$host}{dhcpHostIP}
952 ? ", dhcp=$Jobs{$host}{dhcpHostIP}" : "",
953 ", share=$Jobs{$host}{shareName})\n");
954 $Status{$host}{state} = "Status_backup_in_progress";
955 $Status{$host}{reason} = "";
956 $Status{$host}{type} = $1;
957 $Status{$host}{startTime} = time;
958 $Status{$host}{deadCnt} = 0;
959 $Status{$host}{aliveCnt}++;
960 $Status{$host}{dhcpCheckCnt}--
961 if ( $Status{$host}{dhcpCheckCnt} > 0 );
962 } elsif ( $mesg =~ /^xferPids (.*)/ ) {
963 $Jobs{$host}{xferPid} = $1;
964 } elsif ( $mesg =~ /^completionPercent (.*)/ ) {
965 $Jobs{$host}{completionPercent} = $1;
966 } elsif ( $mesg =~ /^started_restore/ ) {
967 $Jobs{$host}{type} = "restore";
968 print(LOG $bpc->timeStamp,
969 "Started restore on $host"
970 . " (pid=$Jobs{$host}{pid})\n");
971 $Status{$host}{state} = "Status_restore_in_progress";
972 $Status{$host}{reason} = "";
973 $Status{$host}{type} = "restore";
974 $Status{$host}{startTime} = time;
975 $Status{$host}{deadCnt} = 0;
976 $Status{$host}{aliveCnt}++;
977 } elsif ( $mesg =~ /^started_archive/ ) {
978 $Jobs{$host}{type} = "archive";
979 print(LOG $bpc->timeStamp,
980 "Started archive on $host"
981 . " (pid=$Jobs{$host}{pid})\n");
982 $Status{$host}{state} = "Status_archive_in_progress";
983 $Status{$host}{reason} = "";
984 $Status{$host}{type} = "archive";
985 $Status{$host}{startTime} = time;
986 $Status{$host}{deadCnt} = 0;
987 $Status{$host}{aliveCnt}++;
988 } elsif ( $mesg =~ /^(full|incr) backup complete/ ) {
989 print(LOG $bpc->timeStamp, "Finished $1 backup on $host\n");
990 $Status{$host}{reason} = "Reason_backup_done";
991 delete($Status{$host}{error});
992 delete($Status{$host}{errorTime});
993 $Status{$host}{endTime} = time;
994 $Status{$host}{lastGoodBackupTime} = time;
995 } elsif ( $mesg =~ /^backups disabled/ ) {
996 print(LOG $bpc->timeStamp,
997 "Ignoring old backup error on $host\n");
998 $Status{$host}{reason} = "Reason_backup_done";
999 delete($Status{$host}{error});
1000 delete($Status{$host}{errorTime});
1001 $Status{$host}{endTime} = time;
1002 } elsif ( $mesg =~ /^restore complete/ ) {
1003 print(LOG $bpc->timeStamp, "Finished restore on $host\n");
1004 $Status{$host}{reason} = "Reason_restore_done";
1005 delete($Status{$host}{error});
1006 delete($Status{$host}{errorTime});
1007 $Status{$host}{endTime} = time;
1008 } elsif ( $mesg =~ /^archive complete/ ) {
1009 print(LOG $bpc->timeStamp, "Finished archive on $host\n");
1010 $Status{$host}{reason} = "Reason_archive_done";
1011 delete($Status{$host}{error});
1012 delete($Status{$host}{errorTime});
1013 $Status{$host}{endTime} = time;
1014 } elsif ( $mesg =~ /^nothing to do/ ) {
1015 if ( $Status{$host}{reason} ne "Reason_backup_failed"
1016 && $Status{$host}{reason} ne "Reason_restore_failed" ) {
1017 $Status{$host}{state} = "Status_idle";
1018 $Status{$host}{reason} = "Reason_nothing_to_do";
1019 $Status{$host}{startTime} = time;
1021 $Status{$host}{dhcpCheckCnt}--
1022 if ( $Status{$host}{dhcpCheckCnt} > 0 );
1023 } elsif ( $mesg =~ /^no ping response/
1024 || $mesg =~ /^ping too slow/
1025 || $mesg =~ /^host not found/ ) {
1026 $Status{$host}{state} = "Status_idle";
1027 if ( $Status{$host}{userReq}
1028 || $Status{$host}{reason} ne "Reason_backup_failed"
1029 || $Status{$host}{error} =~ /^aborted by user/ ) {
1030 $Status{$host}{reason} = "Reason_no_ping";
1031 $Status{$host}{error} = $mesg;
1032 $Status{$host}{startTime} = time;
1034 $Status{$host}{deadCnt}++;
1035 if ( $Status{$host}{deadCnt} >= $Conf{BlackoutBadPingLimit} ) {
1036 $Status{$host}{aliveCnt} = 0;
1038 } elsif ( $mesg =~ /^dump failed: (.*)/ ) {
1039 $Status{$host}{state} = "Status_idle";
1040 $Status{$host}{error} = $1;
1041 $Status{$host}{errorTime} = time;
1042 $Status{$host}{endTime} = time;
1043 if ( $Status{$host}{reason}
1044 eq "Reason_backup_canceled_by_user" ) {
1045 print(LOG $bpc->timeStamp,
1046 "Backup canceled on $host ($1)\n");
1048 $Status{$host}{reason} = "Reason_backup_failed";
1049 print(LOG $bpc->timeStamp,
1050 "Backup failed on $host ($1)\n");
1052 } elsif ( $mesg =~ /^restore failed: (.*)/ ) {
1053 $Status{$host}{state} = "Status_idle";
1054 $Status{$host}{error} = $1;
1055 $Status{$host}{errorTime} = time;
1056 $Status{$host}{endTime} = time;
1057 if ( $Status{$host}{reason}
1058 eq "Reason_restore_canceled_by_user" ) {
1059 print(LOG $bpc->timeStamp,
1060 "Restore canceled on $host ($1)\n");
1062 $Status{$host}{reason} = "Reason_restore_failed";
1063 print(LOG $bpc->timeStamp,
1064 "Restore failed on $host ($1)\n");
1066 } elsif ( $mesg =~ /^archive failed: (.*)/ ) {
1067 $Status{$host}{state} = "Status_idle";
1068 $Status{$host}{error} = $1;
1069 $Status{$host}{errorTime} = time;
1070 $Status{$host}{endTime} = time;
1071 if ( $Status{$host}{reason}
1072 eq "Reason_archive_canceled_by_user" ) {
1073 print(LOG $bpc->timeStamp,
1074 "Archive canceled on $host ($1)\n");
1076 $Status{$host}{reason} = "Reason_archive_failed";
1077 print(LOG $bpc->timeStamp,
1078 "Archive failed on $host ($1)\n");
1080 } elsif ( $mesg =~ /^log\s+(.*)/ ) {
1081 print(LOG $bpc->timeStamp, "$1\n");
1082 } elsif ( $mesg =~ /^BackupPC_stats (\d+) = (.*)/ ) {
1083 my $chunk = int($1 / 16);
1084 my @f = split(/,/, $2);
1085 $Info{pool}{$f[0]}[$chunk]{FileCnt} += $f[1];
1086 $Info{pool}{$f[0]}[$chunk]{DirCnt} += $f[2];
1087 $Info{pool}{$f[0]}[$chunk]{Kb} += $f[3];
1088 $Info{pool}{$f[0]}[$chunk]{Kb2} += $f[4];
1089 $Info{pool}{$f[0]}[$chunk]{KbRm} += $f[5];
1090 $Info{pool}{$f[0]}[$chunk]{FileCntRm} += $f[6];
1091 $Info{pool}{$f[0]}[$chunk]{FileCntRep} += $f[7];
1092 $Info{pool}{$f[0]}[$chunk]{FileRepMax} = $f[8]
1093 if ( $Info{pool}{$f[0]}[$chunk]{FileRepMax} < $f[8] );
1094 $Info{pool}{$f[0]}[$chunk]{FileCntRename} += $f[9];
1095 $Info{pool}{$f[0]}[$chunk]{FileLinkMax} = $f[10]
1096 if ( $Info{pool}{$f[0]}[$chunk]{FileLinkMax} < $f[10] );
1097 $Info{pool}{$f[0]}[$chunk]{FileLinkTotal} += $f[11];
1098 $Info{pool}{$f[0]}[$chunk]{Time} = time;
1099 } elsif ( $mesg =~ /^BackupPC_nightly lock_off/ ) {
1100 $BackupPCNightlyLock--;
1101 if ( $BackupPCNightlyLock == 0 ) {
1103 # This means the last BackupPC_nightly is done with
1104 # the pool clean, so it's ok to start running regular
1105 # backups again. But starting in 3.0 regular jobs
1106 # are decoupled from BackupPC_nightly.
1108 $RunNightlyWhenIdle = 0;
1110 } elsif ( $mesg =~ /^processState\s+(.+)/ ) {
1111 $Jobs{$host}{processState} = $1;
1112 } elsif ( $mesg =~ /^link\s+(.+)/ ) {
1114 $Status{$h}{needLink} = 1;
1116 print(LOG $bpc->timeStamp, "$host: $mesg\n");
1120 # shut down the client connection if we read EOF
1122 if ( $nbytes <= 0 ) {
1123 close($Jobs{$host}{fh});
1124 vec($FDread, $Jobs{$host}{fn}, 1) = 0;
1125 if ( $CmdJob eq $host || $bpc->isAdminJob($host) ) {
1126 my $cmd = $Jobs{$host}{cmd};
1127 $cmd =~ s/$BinDir\///g;
1128 print(LOG $bpc->timeStamp, "Finished $host ($cmd)\n");
1129 $Status{$host}{state} = "Status_idle";
1130 $Status{$host}{endTime} = time;
1131 if ( $cmd =~ /^BackupPC_nightly\s/ ) {
1132 $BackupPCNightlyJobs--;
1133 #print(LOG $bpc->timeStamp, "BackupPC_nightly done; now"
1134 # . " have $BackupPCNightlyJobs running\n");
1135 if ( $BackupPCNightlyJobs <= 0 ) {
1137 # Last BackupPC_nightly has finished
1139 $BackupPCNightlyJobs = 0;
1140 $RunNightlyWhenIdle = 0;
1143 # Combine the 16 per-directory results
1145 for my $p ( qw(pool cpool) ) {
1146 $Info{"${p}FileCnt"} = 0;
1147 $Info{"${p}DirCnt"} = 0;
1148 $Info{"${p}Kb"} = 0;
1149 $Info{"${p}Kb2"} = 0;
1150 $Info{"${p}KbRm"} = 0;
1151 $Info{"${p}FileCntRm"} = 0;
1152 $Info{"${p}FileCntRep"} = 0;
1153 $Info{"${p}FileRepMax"} = 0;
1154 $Info{"${p}FileCntRename"} = 0;
1155 $Info{"${p}FileLinkMax"} = 0;
1156 $Info{"${p}Time"} = 0;
1157 for ( my $i = 0 ; $i < 16 ; $i++ ) {
1158 $Info{"${p}FileCnt"}
1159 += $Info{pool}{$p}[$i]{FileCnt};
1161 += $Info{pool}{$p}[$i]{DirCnt};
1163 += $Info{pool}{$p}[$i]{Kb};
1165 += $Info{pool}{$p}[$i]{Kb2};
1167 += $Info{pool}{$p}[$i]{KbRm};
1168 $Info{"${p}FileCntRm"}
1169 += $Info{pool}{$p}[$i]{FileCntRm};
1170 $Info{"${p}FileCntRep"}
1171 += $Info{pool}{$p}[$i]{FileCntRep};
1172 $Info{"${p}FileRepMax"}
1173 = $Info{pool}{$p}[$i]{FileRepMax}
1174 if ( $Info{"${p}FileRepMax"} <
1175 $Info{pool}{$p}[$i]{FileRepMax} );
1176 $Info{"${p}FileCntRename"}
1177 += $Info{pool}{$p}[$i]{FileCntRename};
1178 $Info{"${p}FileLinkMax"}
1179 = $Info{pool}{$p}[$i]{FileLinkMax}
1180 if ( $Info{"${p}FileLinkMax"} <
1181 $Info{pool}{$p}[$i]{FileLinkMax} );
1182 $Info{"${p}Time"} = $Info{pool}{$p}[$i]{Time}
1183 if ( $Info{"${p}Time"} <
1184 $Info{pool}{$p}[$i]{Time} );
1186 printf(LOG "%s%s nightly clean removed %d files of"
1188 $bpc->timeStamp, ucfirst($p),
1189 $Info{"${p}FileCntRm"},
1190 $Info{"${p}KbRm"} / (1000 * 1024));
1191 printf(LOG "%s%s is %.2fGB, %d files (%d repeated, "
1192 . "%d max chain, %d max links), %d directories\n",
1193 $bpc->timeStamp, ucfirst($p),
1194 $Info{"${p}Kb"} / (1000 * 1024),
1195 $Info{"${p}FileCnt"}, $Info{"${p}FileCntRep"},
1196 $Info{"${p}FileRepMax"},
1197 $Info{"${p}FileLinkMax"}, $Info{"${p}DirCnt"});
1205 # Queue BackupPC_link to complete the backup
1206 # processing for this host.
1208 if ( defined($Status{$host})
1209 && ($Status{$host}{reason} eq "Reason_backup_done"
1210 || $Status{$host}{needLink}) ) {
1212 } elsif ( defined($Status{$host}) ) {
1213 $Status{$host}{state} = "Status_idle";
1216 delete($Jobs{$host});
1217 $Status{$host}{activeJob} = 0 if ( defined($Status{$host}) );
1221 # When we are idle (empty Jobs, CmdQueue, BgQueue, UserQueue) we
1222 # do a pass over %Status updating the deadCnt and aliveCnt for
1223 # DHCP hosts. The reason we need to do this later is we can't
1224 # be sure whether a DHCP host is alive or dead until we have passed
1225 # over all the DHCP pool.
1227 return if ( @CmdQueue || @BgQueue || @UserQueue || keys(%Jobs) > 1 );
1228 foreach my $host ( keys(%Status) ) {
1229 next if ( $Status{$host}{dhcpCheckCnt} <= 0 );
1230 $Status{$host}{deadCnt} += $Status{$host}{dhcpCheckCnt};
1231 $Status{$host}{dhcpCheckCnt} = 0;
1232 if ( $Status{$host}{deadCnt} >= $Conf{BlackoutBadPingLimit} ) {
1233 $Status{$host}{aliveCnt} = 0;
1238 ############################################################################
1239 # Main_Check_Client_Messages($fdRead)
1241 # Check for, and process, any output from our clients. Also checks
1242 # for new connections to our SERVER_UNIX and SERVER_INET sockets.
1243 ############################################################################
1244 sub Main_Check_Client_Messages
1247 foreach my $client ( keys(%Clients) ) {
1248 next if ( !vec($fdRead, $Clients{$client}{fn}, 1) );
1251 # do a last check to make sure there is something to read so
1252 # we are absolutely sure we won't block.
1254 vec(my $readMask, $Clients{$client}{fn}, 1) = 1;
1255 if ( !select($readMask, undef, undef, 0.0) ) {
1256 print(LOG $bpc->timeStamp, "Botch in Main_Check_Client_Messages:"
1257 . " nothing to read from $client. Debug dump:\n");
1258 my($dump) = Data::Dumper->new(
1259 [ \%Clients, \%Jobs, \$FDread, \$fdRead],
1260 [qw(*Clients, *Jobs *FDread, *fdRead)]);
1262 print(LOG $dump->Dump);
1265 my $nbytes = sysread($Clients{$client}{fh}, $mesg, 1024);
1266 $Clients{$client}{mesg} .= $mesg if ( $nbytes > 0 );
1268 # Process any complete lines received from this client.
1270 while ( $Clients{$client}{mesg} =~ /(.*?)[\n\r]+(.*)/s ) {
1273 $Clients{$client}{mesg} = $2;
1275 # Authenticate the message by checking the MD5 digest
1277 my $md5 = Digest::MD5->new;
1278 if ( $cmd !~ /^(.{22}) (.*)/
1279 || ($md5->add($Clients{$client}{seed}
1280 . $Clients{$client}{mesgCnt}
1281 . $Conf{ServerMesgSecret} . $2),
1282 $md5->b64digest ne $1) ) {
1283 print(LOG $bpc->timeStamp, "Corrupted message '$cmd' from"
1284 . " client '$Clients{$client}{clientName}':"
1285 . " shutting down client connection\n");
1289 $Clients{$client}{mesgCnt}++;
1290 $cmd = decode_utf8($2);
1291 if ( $cmd =~ /^stop (\S+)\s+(\S+)\s+(\S*)/ ) {
1295 $host = $bpc->uriUnesc($host);
1296 if ( $CmdJob ne $host && defined($Status{$host})
1297 && defined($Jobs{$host}) ) {
1298 print(LOG $bpc->timeStamp,
1299 "Stopping current $Jobs{$host}{type} of $host,"
1300 . " request by $user (backoff=$backoff)\n");
1301 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
1303 # Don't close the pipe now; wait until the child
1304 # really exits later. Otherwise close() will
1305 # block until the child has exited.
1307 ##vec($FDread, $Jobs{$host}{fn}, 1) = 0;
1308 ##close($Jobs{$host}{fh});
1309 ##delete($Jobs{$host});
1311 $Status{$host}{state} = "Status_idle";
1312 if ( $Jobs{$host}{type} eq "restore" ) {
1313 $Status{$host}{reason}
1314 = "Reason_restore_canceled_by_user";
1315 } elsif ( $Jobs{$host}{type} eq "archive" ) {
1316 $Status{$host}{reason}
1317 = "Reason_archive_canceled_by_user";
1319 $Status{$host}{reason}
1320 = "Reason_backup_canceled_by_user";
1322 $Status{$host}{activeJob} = 0;
1323 $Status{$host}{startTime} = time;
1324 $reply = "ok: $Jobs{$host}{type} of $host canceled";
1325 } elsif ( $BgQueueOn{$host} || $UserQueueOn{$host} ) {
1326 print(LOG $bpc->timeStamp,
1327 "Stopping pending backup of $host,"
1328 . " request by $user (backoff=$backoff)\n");
1329 @BgQueue = grep($_->{host} ne $host, @BgQueue);
1330 @UserQueue = grep($_->{host} ne $host, @UserQueue);
1331 $BgQueueOn{$host} = $UserQueueOn{$host} = 0;
1332 $reply = "ok: pending backup of $host canceled";
1334 print(LOG $bpc->timeStamp,
1335 "Nothing to do for stop backup of $host,"
1336 . " request by $user (backoff=$backoff)\n");
1337 $reply = "ok: no backup was pending or running";
1339 if ( defined($Status{$host}) && $backoff ne "" ) {
1340 if ( $backoff > 0 ) {
1341 $Status{$host}{backoffTime} = time + $backoff * 3600;
1343 delete($Status{$host}{backoffTime});
1346 } elsif ( $cmd =~ /^backup all$/ ) {
1348 } elsif ( $cmd =~ /^BackupPC_nightly run$/ ) {
1349 $RunNightlyWhenIdle = 1;
1350 } elsif ( $cmd =~ /^queue (\S+)$/ ) {
1352 $host = $bpc->uriUnesc($host);
1353 if ( !defined($Hosts->{$host}) ) {
1354 print(LOG $bpc->timeStamp,
1355 "User requested backup of unknown host $host\n");
1356 $reply = "error: unknown host $host";
1358 if ( QueueOnePC($host) ) {
1359 print(LOG $bpc->timeStamp,
1360 "Disk too full ($Info{DUlastValue}%); skipped 1 host\n");
1361 $Info{DUDailySkipHostCnt}++;
1362 $reply = "error: disk too full to queue $host";
1364 print(LOG $bpc->timeStamp, "Host $host queued by user.\n");
1365 $reply = "ok: $host queued";
1368 } elsif ( $cmd =~ /^backup (\S+)\s+(\S+)\s+(\S+)\s+(\S+)/ ) {
1373 $host = $bpc->uriUnesc($host);
1374 $hostIP = $bpc->uriUnesc($hostIP);
1375 if ( !defined($Hosts->{$host}) ) {
1376 print(LOG $bpc->timeStamp,
1377 "User $user requested backup of unknown host"
1379 $reply = "error: unknown host $host";
1381 print(LOG $bpc->timeStamp,
1382 "User $user requested backup of $host"
1384 if ( $BgQueueOn{$hostIP} ) {
1385 @BgQueue = grep($_->{host} ne $hostIP, @BgQueue);
1386 $BgQueueOn{$hostIP} = 0;
1388 if ( $UserQueueOn{$hostIP} ) {
1389 @UserQueue = grep($_->{host} ne $hostIP, @UserQueue);
1390 $UserQueueOn{$hostIP} = 0;
1392 unshift(@UserQueue, {
1398 dhcp => $hostIP eq $host ? 0 : 1,
1400 $UserQueueOn{$hostIP} = 1;
1401 $reply = "ok: requested backup of $host";
1403 } elsif ( $cmd =~ /^archive (\S+)\s+(\S+)\s+(\S+)/ ) {
1405 my $archivehost = $2;
1406 my $reqFileName = $3;
1407 $host = $bpc->uriUnesc($archivehost);
1408 if ( !defined($Status{$host}) ) {
1409 print(LOG $bpc->timeStamp,
1410 "User $user requested archive of unknown archive host"
1412 $reply = "archive error: unknown archive host $host";
1414 print(LOG $bpc->timeStamp,
1415 "User $user requested archive on $host"
1417 if ( defined($Jobs{$host}) ) {
1418 $reply = "Archive currently running on $host, please try later";
1420 unshift(@UserQueue, {
1423 reqFileName => $reqFileName,
1429 $UserQueueOn{$host} = 1;
1430 $reply = "ok: requested archive on $host";
1433 } elsif ( $cmd =~ /^restore (\S+)\s+(\S+)\s+(\S+)\s+(\S+)/ ) {
1437 my $reqFileName = $4;
1438 $host = $bpc->uriUnesc($host);
1439 $hostIP = $bpc->uriUnesc($hostIP);
1440 if ( !defined($Hosts->{$host}) ) {
1441 print(LOG $bpc->timeStamp,
1442 "User $user requested restore to unknown host"
1444 $reply = "restore error: unknown host $host";
1446 print(LOG $bpc->timeStamp,
1447 "User $user requested restore to $host"
1449 unshift(@UserQueue, {
1452 reqFileName => $reqFileName,
1458 $UserQueueOn{$host} = 1;
1459 if ( defined($Jobs{$host}) ) {
1460 $reply = "ok: requested restore of $host, but a"
1461 . " job is currently running,"
1462 . " so this request will start later";
1464 $reply = "ok: requested restore of $host";
1467 } elsif ( $cmd =~ /^status\s*(.*)/ ) {
1469 my($dump, @values, @names);
1470 foreach my $type ( split(/\s+/, $args) ) {
1471 if ( $type =~ /^queues/ ) {
1472 push(@values, \@BgQueue, \@UserQueue, \@CmdQueue);
1473 push(@names, qw(*BgQueue *UserQueue *CmdQueue));
1474 } elsif ( $type =~ /^jobs/ ) {
1475 push(@values, \%Jobs);
1476 push(@names, qw(*Jobs));
1477 } elsif ( $type =~ /^queueLen/ ) {
1479 BgQueue => scalar(@BgQueue),
1480 UserQueue => scalar(@UserQueue),
1481 CmdQueue => scalar(@CmdQueue),
1483 push(@names, qw(*QueueLen));
1484 } elsif ( $type =~ /^info/ ) {
1485 push(@values, \%Info);
1486 push(@names, qw(*Info));
1487 } elsif ( $type =~ /^hosts/ ) {
1488 push(@values, \%Status);
1489 push(@names, qw(*Status));
1490 } elsif ( $type =~ /^host\((.*)\)/ ) {
1491 my $h = $bpc->uriUnesc($1);
1492 if ( defined($Status{$h}) ) {
1495 BgQueueOn => $BgQueueOn{$h},
1496 UserQueueOn => $UserQueueOn{$h},
1497 CmdQueueOn => $CmdQueueOn{$h},
1499 push(@names, qw(*StatusHost));
1501 print(LOG $bpc->timeStamp,
1502 "Unknown host $h for status request\n");
1505 print(LOG $bpc->timeStamp,
1506 "Unknown status request $type\n");
1509 $dump = Data::Dumper->new(\@values, \@names);
1511 $reply = $dump->Dump;
1512 } elsif ( $cmd =~ /^link\s+(.+)/ ) {
1514 $host = $bpc->uriUnesc($host);
1516 } elsif ( $cmd =~ /^log\s+(.*)/ ) {
1517 print(LOG $bpc->timeStamp, "$1\n");
1518 } elsif ( $cmd =~ /^server\s+(\w+)/ ) {
1520 if ( $type eq 'reload' ) {
1521 ServerReload("Reloading config/host files via CGI request");
1522 } elsif ( $type eq 'shutdown' ) {
1523 $reply = "Shutting down...\n";
1524 syswrite($Clients{$client}{fh}, $reply, length($reply));
1525 ServerShutdown("Server shutting down...");
1527 } elsif ( $cmd =~ /^quit/ || $cmd =~ /^exit/ ) {
1531 print(LOG $bpc->timeStamp, "Unknown command $cmd\n");
1532 $reply = "error: bad command $cmd";
1535 # send a reply to the client, at a minimum "ok\n".
1537 $reply = "ok" if ( $reply eq "" );
1539 syswrite($Clients{$client}{fh}, $reply, length($reply));
1542 # Detect possible denial-of-service attack from sending a huge line
1543 # (ie: never terminated). 32K seems to be plenty big enough as
1546 if ( length($Clients{$client}{mesg}) > 32 * 1024 ) {
1547 print(LOG $bpc->timeStamp, "Line too long from client"
1548 . " '$Clients{$client}{clientName}':"
1549 . " shutting down client connection\n");
1553 # Shut down the client connection if we read EOF
1555 if ( $nbytes <= 0 ) {
1556 close($Clients{$client}{fh});
1557 vec($FDread, $Clients{$client}{fn}, 1) = 0;
1558 delete($Clients{$client});
1562 # Accept any new connections on each of our listen sockets
1564 if ( vec($fdRead, fileno(SERVER_UNIX), 1) ) {
1566 my $paddr = accept(CLIENT, SERVER_UNIX);
1568 $Clients{$ClientConnCnt}{clientName} = "unix socket";
1569 $Clients{$ClientConnCnt}{mesg} = "";
1570 $Clients{$ClientConnCnt}{fh} = *CLIENT;
1571 $Clients{$ClientConnCnt}{fn} = fileno(CLIENT);
1572 vec($FDread, $Clients{$ClientConnCnt}{fn}, 1) = 1;
1574 # Generate and send unique seed for MD5 digests to avoid
1575 # replay attacks. See BackupPC::Lib::ServerMesg().
1577 my $seed = time . ",$ClientConnCnt,$$,0\n";
1578 $Clients{$ClientConnCnt}{seed} = $seed;
1579 $Clients{$ClientConnCnt}{mesgCnt} = 0;
1580 syswrite($Clients{$ClientConnCnt}{fh}, $seed, length($seed));
1582 if ( $ServerInetPort > 0 && vec($fdRead, fileno(SERVER_INET), 1) ) {
1584 my $paddr = accept(CLIENT, SERVER_INET);
1585 my($port,$iaddr) = sockaddr_in($paddr);
1586 my $name = gethostbyaddr($iaddr, AF_INET);
1588 $Clients{$ClientConnCnt}{mesg} = "";
1589 $Clients{$ClientConnCnt}{fh} = *CLIENT;
1590 $Clients{$ClientConnCnt}{fn} = fileno(CLIENT);
1591 $Clients{$ClientConnCnt}{clientName} = "$name:$port";
1592 vec($FDread, $Clients{$ClientConnCnt}{fn}, 1) = 1;
1594 # Generate and send unique seed for MD5 digests to avoid
1595 # replay attacks. See BackupPC::Lib::ServerMesg().
1597 my $seed = time . ",$ClientConnCnt,$$,$port\n";
1598 $Clients{$ClientConnCnt}{seed} = $seed;
1599 $Clients{$ClientConnCnt}{mesgCnt} = 0;
1600 syswrite($Clients{$ClientConnCnt}{fh}, $seed, length($seed));
1604 ###########################################################################
1605 # Miscellaneous subroutines
1606 ###########################################################################
1609 # Write the current status to $LogDir/status.pl
1613 my($dump) = Data::Dumper->new(
1614 [ \%Info, \%Status],
1615 [qw(*Info *Status)]);
1617 my $text = $dump->Dump;
1618 $bpc->{storage}->TextFileWrite("$LogDir/status.pl", $text);
1622 # Compare function for host sort. Hosts with errors go first,
1623 # sorted with the oldest errors first. The remaining hosts
1624 # are sorted so that those with the oldest backups go first.
1629 # Hosts with errors go before hosts without errors
1631 return -1 if ( $Status{$a}{error} ne "" && $Status{$b}{error} eq "" );
1634 # Hosts with no errors go after hosts with errors
1636 return 1 if ( $Status{$a}{error} eq "" && $Status{$b}{error} ne "" );
1639 # hosts with the older last good backups sort earlier
1641 my $r = $Status{$a}{lastGoodBackupTime} <=> $Status{$b}{lastGoodBackupTime};
1642 return $r if ( $r );
1645 # Finally, just sort based on host name
1655 delete($Status{$host}{backoffTime})
1656 if ( defined($Status{$host}{backoffTime})
1657 && $Status{$host}{backoffTime} < time );
1658 return if ( defined($Jobs{$host})
1659 || $BgQueueOn{$host}
1660 || $UserQueueOn{$host}
1661 || $CmdQueueOn{$host} );
1662 if ( $Hosts->{$host}{dhcp} ) {
1663 $Status{$host}{dhcpCheckCnt}++;
1664 if ( $RunNightlyWhenIdle ) {
1666 # Once per night queue a check for DHCP hosts that just
1667 # checks for expired dumps. We need to do this to handle
1668 # the case when a DHCP host has not been on the network for
1669 # a long time, and some of the old dumps need to be expired.
1670 # Normally expiry checks are done by BackupPC_dump only
1671 # after the DHCP hosts has been detected on the network.
1674 {host => $host, user => "BackupPC", reqTime => time,
1675 dhcp => 0, dumpExpire => 1});
1676 $BgQueueOn{$host} = 1;
1680 # this is a fixed ip host: queue it
1682 if ( $Info{DUlastValue} > $Conf{DfMaxUsagePct} ) {
1684 # Since we are out of disk space, instead of queuing
1685 # a regular job, queue an expire check instead. That
1686 # way if the admin reduces the number of backups to
1687 # keep then we will actually delete them. Otherwise
1688 # BackupPC_dump will never run since we have exceeded
1693 {host => $host, user => "BackupPC", reqTime => time,
1694 dhcp => $Hosts->{$host}{dhcp}, dumpExpire => 1});
1697 # Queue regular background backup
1700 {host => $host, user => "BackupPC", reqTime => time,
1701 dhcp => $Hosts->{$host}{dhcp}});
1703 $BgQueueOn{$host} = 1;
1710 # Queue all the hosts for backup. This means queuing all the fixed
1711 # ip hosts and all the dhcp address ranges. We also additionally
1712 # queue the dhcp hosts with a -e flag to check for expired dumps.
1718 foreach my $host ( sort(HostSortCompare keys(%$Hosts)) ) {
1719 $nSkip += QueueOnePC($host);
1722 print(LOG $bpc->timeStamp,
1723 "Disk too full ($Info{DUlastValue}%); skipped $nSkip hosts\n");
1724 $Info{DUDailySkipHostCnt} += $nSkip;
1726 foreach my $dhcp ( @{$Conf{DHCPAddressRanges}} ) {
1727 for ( my $i = $dhcp->{first} ; $i <= $dhcp->{last} ; $i++ ) {
1728 my $ipAddr = "$dhcp->{ipAddrBase}.$i";
1729 next if ( defined($Jobs{$ipAddr})
1730 || $BgQueueOn{$ipAddr}
1731 || $UserQueueOn{$ipAddr}
1732 || $CmdQueueOn{$ipAddr} );
1734 # this is a potential dhcp ip address (we don't know the
1735 # host name yet): queue it
1738 {host => $ipAddr, user => "BackupPC", reqTime => time,
1740 $BgQueueOn{$ipAddr} = 1;
1746 # Queue a BackupPC_link for the given host
1752 return if ( $CmdQueueOn{$host} );
1753 $Status{$host}{state} = "Status_link_pending";
1754 $Status{$host}{needLink} = 0;
1755 unshift(@CmdQueue, {
1759 cmd => ["$BinDir/BackupPC_link", $host],
1761 $CmdQueueOn{$host} = 1;
1765 # Read the hosts file, and update Status if any hosts have been
1766 # added or deleted. We also track the mtime so the only need to
1767 # update the hosts file on changes.
1769 # This function is called at startup, SIGHUP, and on each wakeup.
1770 # It returns 1 on success and undef on failure.
1777 # Nothing to do if we already have the current hosts file
1779 return 1 if ( !$force && defined($Hosts)
1780 && $Info{HostsModTime} == $bpc->HostsMTime() );
1781 if ( !defined($newHosts = $bpc->HostInfoRead()) ) {
1782 print(LOG $bpc->timeStamp, "Can't read hosts file!\n");
1785 print(LOG $bpc->timeStamp, "Reading hosts file\n");
1787 $Info{HostsModTime} = $bpc->HostsMTime();
1789 # Now update %Status in case any hosts have been added or deleted
1791 foreach my $host ( sort(keys(%$Hosts)) ) {
1792 next if ( defined($Status{$host}) );
1793 $Status{$host}{state} = "Status_idle";
1794 print(LOG $bpc->timeStamp, "Added host $host to backup list\n");
1796 foreach my $host ( sort(keys(%Status)) ) {
1797 next if ( $host eq $bpc->trashJob
1798 || $bpc->isAdminJob($host)
1799 || defined($Hosts->{$host})
1800 || defined($Jobs{$host})
1801 || $BgQueueOn{$host}
1802 || $UserQueueOn{$host}
1803 || $CmdQueueOn{$host} );
1804 print(LOG $bpc->timeStamp, "Deleted host $host from backup list\n");
1805 delete($Status{$host});
1811 # Remember the signal name for later processing
1817 foreach my $host ( keys(%Jobs) ) {
1818 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
1821 # In case we are inside the exit handler, reopen the log file
1825 print(LOG "Fatal error: unhandled signal $SigName\n");
1826 unlink("$LogDir/BackupPC.pid");
1827 confess("Got new signal $SigName... quitting\n");
1834 # Open the log file and point STDOUT and STDERR there too
1838 mkpath($LogDir, 0, 0777) if ( !-d $LogDir );
1839 open(LOG, ">>$LogDir/LOG")
1840 || die("Can't create LOG file $LogDir/LOG");
1843 open(STDOUT, ">&LOG");
1844 open(STDERR, ">&LOG");
1845 select(LOG); $| = 1;
1846 select(STDERR); $| = 1;
1847 select(STDOUT); $| = 1;
1851 # Initialize the unix-domain and internet-domain sockets that
1852 # we listen to for client connections (from the CGI script and
1853 # some of the BackupPC sub-programs).
1855 sub ServerSocketInit
1857 if ( !defined(fileno(SERVER_UNIX)) ) {
1859 # one-time only: initialize unix-domain socket
1861 if ( !socket(SERVER_UNIX, PF_UNIX, SOCK_STREAM, 0) ) {
1862 print(LOG $bpc->timeStamp, "unix socket() failed: $!\n");
1865 my $sockFile = "$LogDir/BackupPC.sock";
1867 if ( !bind(SERVER_UNIX, sockaddr_un($sockFile)) ) {
1868 print(LOG $bpc->timeStamp, "unix bind() failed: $!\n");
1871 if ( !listen(SERVER_UNIX, SOMAXCONN) ) {
1872 print(LOG $bpc->timeStamp, "unix listen() failed: $!\n");
1875 vec($FDread, fileno(SERVER_UNIX), 1) = 1;
1877 return if ( $ServerInetPort == $Conf{ServerPort} );
1878 if ( $ServerInetPort > 0 ) {
1879 vec($FDread, fileno(SERVER_INET), 1) = 0;
1881 $ServerInetPort = -1;
1883 if ( $Conf{ServerPort} > 0 ) {
1885 # Setup a socket to listen on $Conf{ServerPort}
1887 my $proto = getprotobyname('tcp');
1888 if ( !socket(SERVER_INET, PF_INET, SOCK_STREAM, $proto) ) {
1889 print(LOG $bpc->timeStamp, "inet socket() failed: $!\n");
1892 if ( !setsockopt(SERVER_INET, SOL_SOCKET, SO_REUSEADDR, pack("l",1)) ) {
1893 print(LOG $bpc->timeStamp, "setsockopt() failed: $!\n");
1896 if ( !bind(SERVER_INET, sockaddr_in($Conf{ServerPort}, INADDR_ANY)) ) {
1897 print(LOG $bpc->timeStamp, "inet bind() failed: $!\n");
1900 if ( !listen(SERVER_INET, SOMAXCONN) ) {
1901 print(LOG $bpc->timeStamp, "inet listen() failed: $!\n");
1904 vec($FDread, fileno(SERVER_INET), 1) = 1;
1905 $ServerInetPort = $Conf{ServerPort};
1910 # Reload the server. Used by Main_Process_Signal when $SigName eq "HUP"
1911 # or when the command "server reload" is received.
1916 $mesg = $bpc->ConfigRead() || $mesg;
1917 print(LOG $bpc->timeStamp, "$mesg\n");
1918 $Info{ConfigModTime} = $bpc->ConfigMTime();
1919 %Conf = $bpc->Conf();
1920 umask($Conf{UmaskMode});
1924 $Info{ConfigLTime} = time;
1928 # Gracefully shutdown the server. Used by Main_Process_Signal when
1929 # $SigName ne "" && $SigName ne "HUP" or when the command
1930 # "server shutdown" is received.
1935 print(LOG $bpc->timeStamp, "$mesg\n");
1936 if ( keys(%Jobs) ) {
1937 foreach my $host ( keys(%Jobs) ) {
1938 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
1941 foreach my $host ( keys(%Jobs) ) {
1942 kill($bpc->sigName2num("KILL"), $Jobs{$host}{pid});
1948 unlink("$LogDir/BackupPC.pid");