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 3.2.0, released 31 Dec 2008.
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 ) {
414 # Queue multiple nightly jobs based on the configuration
416 $Conf{MaxBackupPCNightlyJobs} = 1
417 if ( $Conf{MaxBackupPCNightlyJobs} <= 0 );
418 $Conf{BackupPCNightlyPeriod} = 1
419 if ( $Conf{BackupPCNightlyPeriod} <= 0 );
421 # Decide what subset of the 16 top-level directories 0..9a..f
422 # we run BackupPC_nightly on, based on $Conf{BackupPCNightlyPeriod}.
423 # If $Conf{BackupPCNightlyPeriod} == 1 then we run 0..15 every
424 # time. If $Conf{BackupPCNightlyPeriod} == 2 then we run
425 # 0..7 one night and 89a-f the next night. And so on.
427 # $Info{NightlyPhase} counts which night, from 0 to
428 # $Conf{BackupPCNightlyPeriod} - 1.
430 my $start = int($Info{NightlyPhase} * 16
431 / $Conf{BackupPCNightlyPeriod});
432 my $end = int(($Info{NightlyPhase} + 1) * 16
433 / $Conf{BackupPCNightlyPeriod});
434 $end = $start + 1 if ( $end <= $start );
435 $Info{NightlyPhase}++;
436 $Info{NightlyPhase} = 0 if ( $end >= 16 );
439 # Zero out the data we expect to get from BackupPC_nightly.
441 for my $p ( qw(pool cpool) ) {
442 for ( my $i = $start ; $i < $end ; $i++ ) {
443 $Info{pool}{$p}[$i]{FileCnt} = 0;
444 $Info{pool}{$p}[$i]{DirCnt} = 0;
445 $Info{pool}{$p}[$i]{Kb} = 0;
446 $Info{pool}{$p}[$i]{Kb2} = 0;
447 $Info{pool}{$p}[$i]{KbRm} = 0;
448 $Info{pool}{$p}[$i]{FileCntRm} = 0;
449 $Info{pool}{$p}[$i]{FileCntRep} = 0;
450 $Info{pool}{$p}[$i]{FileRepMax} = 0;
451 $Info{pool}{$p}[$i]{FileCntRename} = 0;
452 $Info{pool}{$p}[$i]{FileLinkMax} = 0;
453 $Info{pool}{$p}[$i]{Time} = 0;
456 print(LOG $bpc->timeStamp,
457 sprintf("Running %d BackupPC_nightly jobs from %d..%d"
458 . " (out of 0..15)\n",
459 $Conf{MaxBackupPCNightlyJobs}, $start, $end - 1));
462 # Now queue the $Conf{MaxBackupPCNightlyJobs} jobs.
463 # The granularity on start and end is now 0..255.
468 for ( my $i = 0 ; $i < $Conf{MaxBackupPCNightlyJobs} ; $i++ ) {
470 # The first nightly job gets the -m option (does email, log aging).
471 # All jobs get the start and end options from 0..255 telling
472 # them which parts of the pool to traverse.
474 my $cmd = ["$BinDir/BackupPC_nightly"];
475 push(@$cmd, "-m") if ( $i == 0 );
477 $start = $start0 + int(($end - $start0)
478 * ($i + 1) / $Conf{MaxBackupPCNightlyJobs});
479 push(@$cmd, $start - 1);
481 my $job = $bpc->adminJob($i);
488 $CmdQueueOn{$job} = 1;
490 $RunNightlyWhenIdle = 2;
494 ############################################################################
495 # Main_TryToRun_CmdQueue()
497 # Decide if we can run a new command from the @CmdQueue.
498 # We only run one of these at a time. The @CmdQueue is
499 # used to run BackupPC_link (for the corresponding host),
500 # BackupPC_trashClean, and BackupPC_nightly using a fake
501 # host name of $bpc->adminJob.
502 ############################################################################
503 sub Main_TryToRun_CmdQueue
507 while ( $CmdJob eq "" && @CmdQueue > 0 && $RunNightlyWhenIdle != 1
508 || @CmdQueue > 0 && $RunNightlyWhenIdle == 2
509 && $bpc->isAdminJob($CmdQueue[0]->{host})
512 $req = pop(@CmdQueue);
514 $host = $req->{host};
515 if ( defined($Jobs{$host}) ) {
516 print(LOG $bpc->timeStamp,
517 "Botch on admin job for $host: already in use!!\n");
519 # This could happen during normal opertion: a user could
520 # request a backup while a BackupPC_link is queued from
521 # a previous backup. But it is unlikely. Just put this
522 # request back on the end of the queue.
524 unshift(@CmdQueue, $req);
527 $CmdQueueOn{$host} = 0;
528 my $cmd = $req->{cmd};
529 my $pid = open(FH, "-|");
530 if ( !defined($pid) ) {
531 print(LOG $bpc->timeStamp,
532 "can't fork for $host, request by $req->{user}\n");
538 $ENV{BPC_REQUSER} = $req->{user};
539 POSIX::nice($Conf{CmdQueueNice}) if ( $Conf{CmdQueueNice} );
541 print(LOG $bpc->timeStamp, "can't exec @$cmd for $host\n");
544 $Jobs{$host}{pid} = $pid;
545 $Jobs{$host}{fh} = *FH;
546 $Jobs{$host}{fn} = fileno(FH);
547 vec($FDread, $Jobs{$host}{fn}, 1) = 1;
548 $Jobs{$host}{startTime} = time;
549 $Jobs{$host}{reqTime} = $req->{reqTime};
550 $cmd = $bpc->execCmd2ShellCmd(@$cmd);
551 $Jobs{$host}{cmd} = $cmd;
552 $Jobs{$host}{user} = $req->{user};
553 $Jobs{$host}{type} = $Status{$host}{type};
554 $Status{$host}{state} = "Status_link_running";
555 $Status{$host}{activeJob} = 1;
556 $Status{$host}{endTime} = time;
557 $CmdJob = $host if ( $host ne $bpc->trashJob );
558 $cmd =~ s/$BinDir\///g;
559 print(LOG $bpc->timeStamp, "Running $cmd (pid=$pid)\n");
560 if ( $cmd =~ /^BackupPC_nightly\s/ ) {
561 $BackupPCNightlyJobs++;
562 $BackupPCNightlyLock++;
567 ############################################################################
568 # Main_TryToRun_Bg_or_User_Queue()
570 # Decide if we can run any new backup requests from @BgQueue
571 # or @UserQueue. Several of these can be run at the same time
572 # based on %Conf settings. Jobs from @UserQueue take priority,
573 # and at total of $Conf{MaxBackups} + $Conf{MaxUserBackups}
574 # simultaneous jobs can run from @UserQueue. After @UserQueue
575 # is exhausted, up to $Conf{MaxBackups} simultaneous jobs can
577 ############################################################################
578 sub Main_TryToRun_Bg_or_User_Queue
581 my(@deferUserQueue, @deferBgQueue);
584 if ( time - $Info{DUlastValueTime} >= 600 ) {
586 # Update our notion of disk usage no more than
587 # once every 10 minutes
589 $du = $bpc->CheckFileSystemUsage($TopDir);
590 $Info{DUlastValue} = $du;
591 $Info{DUlastValueTime} = time;
594 # if we recently checked it then just use the old value
596 $du = $Info{DUlastValue};
598 if ( $Info{DUDailyMaxReset} ) {
599 $Info{DUDailyMaxStartTime} = time;
600 $Info{DUDailyMaxReset} = 0;
601 $Info{DUDailyMax} = 0;
603 if ( $du > $Info{DUDailyMax} ) {
604 $Info{DUDailyMax} = $du;
605 $Info{DUDailyMaxTime} = time;
607 if ( $du > $Conf{DfMaxUsagePct} ) {
608 my @bgQueue = @BgQueue;
612 # When the disk is too full, only run backups that will
613 # do expires, not regular backups
616 foreach $req ( @bgQueue ) {
617 if ( $req->{dumpExpire} ) {
618 unshift(@BgQueue, $req);
620 $BgQueueOn{$req->{host}} = 0;
625 print(LOG $bpc->timeStamp,
626 "Disk too full ($du%); skipped $nSkip hosts\n");
627 $Info{DUDailySkipHostCnt} += $nSkip;
632 # Run background jobs anytime. Previously they were locked out
633 # when BackupPC_nightly was running or pending with this
634 # condition on the while loop:
636 # while ( $RunNightlyWhenIdle == 0 )
640 my(@args, $progName, $type);
641 my $nJobs = keys(%Jobs);
643 # CmdJob and trashClean don't count towards MaxBackups / MaxUserBackups
645 if ( $CmdJob ne "" ) {
646 if ( $BackupPCNightlyJobs ) {
647 $nJobs -= $BackupPCNightlyJobs;
652 $nJobs-- if ( defined($Jobs{$bpc->trashJob} ) );
653 if ( $nJobs < $Conf{MaxBackups} + $Conf{MaxUserBackups}
654 && @UserQueue > 0 ) {
655 $req = pop(@UserQueue);
656 if ( defined($Jobs{$req->{host}}) ) {
657 push(@deferUserQueue, $req);
660 push(@args, $req->{doFull} ? "-f" : "-i")
661 if (( !$req->{restore} ) && ( !$req->{archive} ));
662 $UserQueueOn{$req->{host}} = 0;
663 } elsif ( $nJobs < $Conf{MaxBackups}
664 && (@CmdQueue + $nJobs)
665 <= $Conf{MaxBackups} + $Conf{MaxPendingCmds}
667 $req = pop(@BgQueue);
668 if ( defined($Jobs{$req->{host}}) ) {
670 # Job is currently running for this host; save it for later
672 unshift(@deferBgQueue, $req);
675 $BgQueueOn{$req->{host}} = 0;
678 # Restore the deferred jobs
680 @BgQueue = (@BgQueue, @deferBgQueue);
681 @UserQueue = (@UserQueue, @deferUserQueue);
684 $host = $req->{host};
685 my $user = $req->{user};
686 if ( $req->{restore} ) {
687 $progName = "BackupPC_restore";
689 push(@args, $req->{hostIP}, $req->{host}, $req->{reqFileName});
690 } elsif ( $req->{archive} ) {
691 $progName = "BackupPC_archive";
693 push(@args, $req->{user}, $req->{host}, $req->{reqFileName});
695 $progName = "BackupPC_dump";
697 push(@args, "-d") if ( $req->{dhcp} );
698 push(@args, "-e") if ( $req->{dumpExpire} );
701 my $pid = open(FH, "-|");
702 if ( !defined($pid) ) {
703 print(LOG $bpc->timeStamp,
704 "can't fork to run $progName for $host, request by $user\n");
710 exec("$BinDir/$progName", @args);
711 print(LOG $bpc->timeStamp, "can't exec $progName for $host\n");
714 $Jobs{$host}{pid} = $pid;
715 $Jobs{$host}{fh} = *FH;
716 $Jobs{$host}{fn} = fileno(FH);
717 $Jobs{$host}{dhcp} = $req->{dhcp};
718 vec($FDread, $Jobs{$host}{fn}, 1) = 1;
719 $Jobs{$host}{startTime} = time;
720 $Jobs{$host}{reqTime} = $req->{reqTime};
721 $Jobs{$host}{userReq} = $req->{userReq};
722 $Jobs{$host}{cmd} = $bpc->execCmd2ShellCmd($progName, @args);
723 $Jobs{$host}{user} = $user;
724 $Jobs{$host}{type} = $type;
725 $Status{$host}{userReq} = $req->{userReq}
726 if ( defined($Hosts->{$host}) );
727 if ( !$req->{dhcp} ) {
728 $Status{$host}{state} = "Status_".$type."_starting";
729 $Status{$host}{activeJob} = 1;
730 $Status{$host}{startTime} = time;
731 $Status{$host}{endTime} = "";
736 ############################################################################
739 # If necessary, figure out when to next wakeup based on $Conf{WakeupSchedule},
740 # and then do a select() to wait for the next thing to happen
741 # (timeout, signal, someone sends a message, child dies etc).
742 ############################################################################
745 if ( $NextWakeup <= 0 ) {
747 # Figure out when to next wakeup based on $Conf{WakeupSchedule}.
749 my($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst)
751 my($currHours) = $hour + $min / 60 + $sec / 3600;
752 if ( $bpc->ConfigMTime() != $Info{ConfigModTime} ) {
753 ServerReload("Re-read config file because mtime changed");
756 foreach my $t ( @{$Conf{WakeupSchedule} || [0..23]} ) {
757 next if ( $t < 0 || $t > 24 );
758 my $tomorrow = $t + 24;
760 || ($tomorrow - $currHours > 0
761 && $delta > $tomorrow - $currHours) ) {
762 $delta = $tomorrow - $currHours;
763 $FirstWakeup = $t == $Conf{WakeupSchedule}[0];
766 || ($t - $currHours > 0 && $delta > $t - $currHours) ) {
767 $delta = $t - $currHours;
768 $FirstWakeup = $t == $Conf{WakeupSchedule}[0];
771 $NextWakeup = time + $delta * 3600;
772 $Info{nextWakeup} = $NextWakeup;
773 print(LOG $bpc->timeStamp, "Next wakeup is ",
774 $bpc->timeStamp($NextWakeup, 1), "\n");
777 # Call select(), waiting until either a signal, a timeout,
778 # any output from our jobs, or any messages from clients
780 # select() is where we (hopefully) spend most of our time blocked...
782 my $timeout = $NextWakeup - time;
783 $timeout = 1 if ( $timeout <= 0 );
785 select(my $rout = $FDread, undef, $ein, $timeout);
790 ############################################################################
791 # Main_Process_Signal()
794 ############################################################################
795 sub Main_Process_Signal
800 if ( $SigName eq "HUP" ) {
801 ServerReload("Re-read config file because of a SIG_HUP");
802 } elsif ( $SigName ) {
803 ServerShutdown("Got signal $SigName... cleaning up");
808 ############################################################################
809 # Main_Check_Timeout()
811 # Check if a timeout has occured, and if so, queue all the PCs for backups.
812 # Also does log file aging on the first timeout after midnight.
813 ############################################################################
814 sub Main_Check_Timeout
819 return if ( time < $NextWakeup || $NextWakeup <= 0 );
821 if ( $FirstWakeup ) {
823 # This is the first wakeup after midnight. Do log file aging
824 # and various house keeping.
827 printf(LOG "%s24hr disk usage: %d%% max, %d%% recent,"
828 . " %d skipped hosts\n",
829 $bpc->timeStamp, $Info{DUDailyMax}, $Info{DUlastValue},
830 $Info{DUDailySkipHostCnt});
831 $Info{DUDailyMaxReset} = 1;
832 $Info{DUDailyMaxPrev} = $Info{DUDailyMax};
833 $Info{DUDailySkipHostCntPrev} = $Info{DUDailySkipHostCnt};
834 $Info{DUDailySkipHostCnt} = 0;
835 my $lastLog = $Conf{MaxOldLogFiles} - 1;
836 if ( -f "$LogDir/LOG.$lastLog" ) {
837 print(LOG $bpc->timeStamp,
838 "Removing $LogDir/LOG.$lastLog\n");
839 unlink("$LogDir/LOG.$lastLog");
841 if ( -f "$LogDir/LOG.$lastLog.z" ) {
842 print(LOG $bpc->timeStamp,
843 "Removing $LogDir/LOG.$lastLog.z\n");
844 unlink("$LogDir/LOG.$lastLog.z");
846 print(LOG $bpc->timeStamp, "Aging LOG files, LOG -> LOG.0 -> "
847 . "LOG.1 -> ... -> LOG.$lastLog\n");
848 close(STDERR); # dup of LOG
849 close(STDOUT); # dup of LOG
851 for ( my $i = $lastLog - 1 ; $i >= 0 ; $i-- ) {
853 rename("$LogDir/LOG.$i", "$LogDir/LOG.$j")
854 if ( -f "$LogDir/LOG.$i" );
855 rename("$LogDir/LOG.$i.z", "$LogDir/LOG.$j.z")
856 if ( -f "$LogDir/LOG.$i.z" );
859 # Compress the log file LOG -> LOG.0.z (if enabled).
860 # Otherwise, just rename LOG -> LOG.0.
862 BackupPC::FileZIO->compressCopy("$LogDir/LOG",
865 $Conf{CompressLevel}, 1);
868 # Remember to run the nightly script when the next CmdQueue
871 $RunNightlyWhenIdle = 1;
874 # Write out the current status and then queue all the PCs
878 %BgQueueOn = () if ( @BgQueue == 0 );
879 %UserQueueOn = () if ( @UserQueue == 0 );
880 %CmdQueueOn = () if ( @CmdQueue == 0 );
884 ############################################################################
885 # Main_Check_Job_Messages($fdRead)
887 # Check if select() says we have bytes waiting from any of our jobs.
888 # Handle each of the messages when complete (newline terminated).
889 ############################################################################
890 sub Main_Check_Job_Messages
893 foreach my $host ( keys(%Jobs) ) {
894 next if ( !vec($fdRead, $Jobs{$host}{fn}, 1) );
897 # do a last check to make sure there is something to read so
898 # we are absolutely sure we won't block.
900 vec(my $readMask, $Jobs{$host}{fn}, 1) = 1;
901 if ( !select($readMask, undef, undef, 0.0) ) {
902 print(LOG $bpc->timeStamp, "Botch in Main_Check_Job_Messages:"
903 . " nothing to read from $host. Debug dump:\n");
904 my($dump) = Data::Dumper->new(
905 [ \%Clients, \%Jobs, \$FDread, \$fdRead],
906 [qw(*Clients, *Jobs *FDread, *fdRead)]);
908 print(LOG $dump->Dump);
911 my $nbytes = sysread($Jobs{$host}{fh}, $mesg, 1024);
912 $Jobs{$host}{mesg} .= $mesg if ( $nbytes > 0 );
914 # Process any complete lines of output from this jobs.
915 # Any output to STDOUT or STDERR from the children is processed here.
917 while ( $Jobs{$host}{mesg} =~ /(.*?)[\n\r]+(.*)/s ) {
919 $Jobs{$host}{mesg} = $2;
920 if ( $Jobs{$host}{dhcp} ) {
921 if ( $mesg =~ /^DHCP (\S+) (\S+)/ ) {
922 my $newHost = $bpc->uriUnesc($2);
923 if ( defined($Jobs{$newHost}) ) {
924 print(LOG $bpc->timeStamp,
925 "Backup on $newHost is already running\n");
926 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
930 $Jobs{$host}{dhcpHostIP} = $host;
931 $Status{$newHost}{dhcpHostIP} = $host;
932 $Jobs{$newHost} = $Jobs{$host};
933 delete($Jobs{$host});
935 $Status{$host}{state} = "Status_backup_starting";
936 $Status{$host}{activeJob} = 1;
937 $Status{$host}{startTime} = $Jobs{$host}{startTime};
938 $Status{$host}{endTime} = "";
939 $Jobs{$host}{dhcp} = 0;
941 print(LOG $bpc->timeStamp, "dhcp $host: $mesg\n");
943 } elsif ( $mesg =~ /^started (.*) dump, share=(.*)/ ) {
944 $Jobs{$host}{type} = $1;
945 $Jobs{$host}{shareName} = $2;
946 print(LOG $bpc->timeStamp,
947 "Started $1 backup on $host (pid=$Jobs{$host}{pid}",
948 $Jobs{$host}{dhcpHostIP}
949 ? ", dhcp=$Jobs{$host}{dhcpHostIP}" : "",
950 ", share=$Jobs{$host}{shareName})\n");
951 $Status{$host}{state} = "Status_backup_in_progress";
952 $Status{$host}{reason} = "";
953 $Status{$host}{type} = $1;
954 $Status{$host}{startTime} = time;
955 $Status{$host}{deadCnt} = 0;
956 $Status{$host}{aliveCnt}++;
957 $Status{$host}{dhcpCheckCnt}--
958 if ( $Status{$host}{dhcpCheckCnt} > 0 );
959 } elsif ( $mesg =~ /^xferPids (.*)/ ) {
960 $Jobs{$host}{xferPid} = $1;
961 } elsif ( $mesg =~ /^started_restore/ ) {
962 $Jobs{$host}{type} = "restore";
963 print(LOG $bpc->timeStamp,
964 "Started restore on $host"
965 . " (pid=$Jobs{$host}{pid})\n");
966 $Status{$host}{state} = "Status_restore_in_progress";
967 $Status{$host}{reason} = "";
968 $Status{$host}{type} = "restore";
969 $Status{$host}{startTime} = time;
970 $Status{$host}{deadCnt} = 0;
971 $Status{$host}{aliveCnt}++;
972 } elsif ( $mesg =~ /^started_archive/ ) {
973 $Jobs{$host}{type} = "archive";
974 print(LOG $bpc->timeStamp,
975 "Started archive on $host"
976 . " (pid=$Jobs{$host}{pid})\n");
977 $Status{$host}{state} = "Status_archive_in_progress";
978 $Status{$host}{reason} = "";
979 $Status{$host}{type} = "archive";
980 $Status{$host}{startTime} = time;
981 $Status{$host}{deadCnt} = 0;
982 $Status{$host}{aliveCnt}++;
983 } elsif ( $mesg =~ /^(full|incr) backup complete/ ) {
984 print(LOG $bpc->timeStamp, "Finished $1 backup on $host\n");
985 $Status{$host}{reason} = "Reason_backup_done";
986 delete($Status{$host}{error});
987 delete($Status{$host}{errorTime});
988 $Status{$host}{endTime} = time;
989 $Status{$host}{lastGoodBackupTime} = time;
990 } elsif ( $mesg =~ /^backups disabled/ ) {
991 print(LOG $bpc->timeStamp,
992 "Ignoring old backup error on $host\n");
993 $Status{$host}{reason} = "Reason_backup_done";
994 delete($Status{$host}{error});
995 delete($Status{$host}{errorTime});
996 $Status{$host}{endTime} = time;
997 } elsif ( $mesg =~ /^restore complete/ ) {
998 print(LOG $bpc->timeStamp, "Finished restore on $host\n");
999 $Status{$host}{reason} = "Reason_restore_done";
1000 delete($Status{$host}{error});
1001 delete($Status{$host}{errorTime});
1002 $Status{$host}{endTime} = time;
1003 } elsif ( $mesg =~ /^archive complete/ ) {
1004 print(LOG $bpc->timeStamp, "Finished archive on $host\n");
1005 $Status{$host}{reason} = "Reason_archive_done";
1006 delete($Status{$host}{error});
1007 delete($Status{$host}{errorTime});
1008 $Status{$host}{endTime} = time;
1009 } elsif ( $mesg =~ /^nothing to do/ ) {
1010 if ( $Status{$host}{reason} ne "Reason_backup_failed"
1011 && $Status{$host}{reason} ne "Reason_restore_failed" ) {
1012 $Status{$host}{state} = "Status_idle";
1013 $Status{$host}{reason} = "Reason_nothing_to_do";
1014 $Status{$host}{startTime} = time;
1016 $Status{$host}{dhcpCheckCnt}--
1017 if ( $Status{$host}{dhcpCheckCnt} > 0 );
1018 } elsif ( $mesg =~ /^no ping response/
1019 || $mesg =~ /^ping too slow/
1020 || $mesg =~ /^host not found/ ) {
1021 $Status{$host}{state} = "Status_idle";
1022 if ( $Status{$host}{userReq}
1023 || $Status{$host}{reason} ne "Reason_backup_failed"
1024 || $Status{$host}{error} =~ /^aborted by user/ ) {
1025 $Status{$host}{reason} = "Reason_no_ping";
1026 $Status{$host}{error} = $mesg;
1027 $Status{$host}{startTime} = time;
1029 $Status{$host}{deadCnt}++;
1030 if ( $Status{$host}{deadCnt} >= $Conf{BlackoutBadPingLimit} ) {
1031 $Status{$host}{aliveCnt} = 0;
1033 } elsif ( $mesg =~ /^dump failed: (.*)/ ) {
1034 $Status{$host}{state} = "Status_idle";
1035 $Status{$host}{error} = $1;
1036 $Status{$host}{errorTime} = time;
1037 $Status{$host}{endTime} = time;
1038 if ( $Status{$host}{reason}
1039 eq "Reason_backup_canceled_by_user" ) {
1040 print(LOG $bpc->timeStamp,
1041 "Backup canceled on $host ($1)\n");
1043 $Status{$host}{reason} = "Reason_backup_failed";
1044 print(LOG $bpc->timeStamp,
1045 "Backup failed on $host ($1)\n");
1047 } elsif ( $mesg =~ /^restore failed: (.*)/ ) {
1048 $Status{$host}{state} = "Status_idle";
1049 $Status{$host}{error} = $1;
1050 $Status{$host}{errorTime} = time;
1051 $Status{$host}{endTime} = time;
1052 if ( $Status{$host}{reason}
1053 eq "Reason_restore_canceled_by_user" ) {
1054 print(LOG $bpc->timeStamp,
1055 "Restore canceled on $host ($1)\n");
1057 $Status{$host}{reason} = "Reason_restore_failed";
1058 print(LOG $bpc->timeStamp,
1059 "Restore failed on $host ($1)\n");
1061 } elsif ( $mesg =~ /^archive failed: (.*)/ ) {
1062 $Status{$host}{state} = "Status_idle";
1063 $Status{$host}{error} = $1;
1064 $Status{$host}{errorTime} = time;
1065 $Status{$host}{endTime} = time;
1066 if ( $Status{$host}{reason}
1067 eq "Reason_archive_canceled_by_user" ) {
1068 print(LOG $bpc->timeStamp,
1069 "Archive canceled on $host ($1)\n");
1071 $Status{$host}{reason} = "Reason_archive_failed";
1072 print(LOG $bpc->timeStamp,
1073 "Archive failed on $host ($1)\n");
1075 } elsif ( $mesg =~ /^log\s+(.*)/ ) {
1076 print(LOG $bpc->timeStamp, "$1\n");
1077 } elsif ( $mesg =~ /^BackupPC_stats (\d+) = (.*)/ ) {
1078 my $chunk = int($1 / 16);
1079 my @f = split(/,/, $2);
1080 $Info{pool}{$f[0]}[$chunk]{FileCnt} += $f[1];
1081 $Info{pool}{$f[0]}[$chunk]{DirCnt} += $f[2];
1082 $Info{pool}{$f[0]}[$chunk]{Kb} += $f[3];
1083 $Info{pool}{$f[0]}[$chunk]{Kb2} += $f[4];
1084 $Info{pool}{$f[0]}[$chunk]{KbRm} += $f[5];
1085 $Info{pool}{$f[0]}[$chunk]{FileCntRm} += $f[6];
1086 $Info{pool}{$f[0]}[$chunk]{FileCntRep} += $f[7];
1087 $Info{pool}{$f[0]}[$chunk]{FileRepMax} = $f[8]
1088 if ( $Info{pool}{$f[0]}[$chunk]{FileRepMax} < $f[8] );
1089 $Info{pool}{$f[0]}[$chunk]{FileCntRename} += $f[9];
1090 $Info{pool}{$f[0]}[$chunk]{FileLinkMax} = $f[10]
1091 if ( $Info{pool}{$f[0]}[$chunk]{FileLinkMax} < $f[10] );
1092 $Info{pool}{$f[0]}[$chunk]{FileLinkTotal} += $f[11];
1093 $Info{pool}{$f[0]}[$chunk]{Time} = time;
1094 } elsif ( $mesg =~ /^BackupPC_nightly lock_off/ ) {
1095 $BackupPCNightlyLock--;
1096 if ( $BackupPCNightlyLock == 0 ) {
1098 # This means the last BackupPC_nightly is done with
1099 # the pool clean, so it's ok to start running regular
1100 # backups again. But starting in 3.0 regular jobs
1101 # are decoupled from BackupPC_nightly.
1103 $RunNightlyWhenIdle = 0;
1105 } elsif ( $mesg =~ /^processState\s+(.+)/ ) {
1106 $Jobs{$host}{processState} = $1;
1107 } elsif ( $mesg =~ /^link\s+(.+)/ ) {
1109 $Status{$h}{needLink} = 1;
1111 print(LOG $bpc->timeStamp, "$host: $mesg\n");
1115 # shut down the client connection if we read EOF
1117 if ( $nbytes <= 0 ) {
1118 close($Jobs{$host}{fh});
1119 vec($FDread, $Jobs{$host}{fn}, 1) = 0;
1120 if ( $CmdJob eq $host || $bpc->isAdminJob($host) ) {
1121 my $cmd = $Jobs{$host}{cmd};
1122 $cmd =~ s/$BinDir\///g;
1123 print(LOG $bpc->timeStamp, "Finished $host ($cmd)\n");
1124 $Status{$host}{state} = "Status_idle";
1125 $Status{$host}{endTime} = time;
1126 if ( $cmd =~ /^BackupPC_nightly\s/ ) {
1127 $BackupPCNightlyJobs--;
1128 #print(LOG $bpc->timeStamp, "BackupPC_nightly done; now"
1129 # . " have $BackupPCNightlyJobs running\n");
1130 if ( $BackupPCNightlyJobs <= 0 ) {
1132 # Last BackupPC_nightly has finished
1134 $BackupPCNightlyJobs = 0;
1135 $RunNightlyWhenIdle = 0;
1138 # Combine the 16 per-directory results
1140 for my $p ( qw(pool cpool) ) {
1141 $Info{"${p}FileCnt"} = 0;
1142 $Info{"${p}DirCnt"} = 0;
1143 $Info{"${p}Kb"} = 0;
1144 $Info{"${p}Kb2"} = 0;
1145 $Info{"${p}KbRm"} = 0;
1146 $Info{"${p}FileCntRm"} = 0;
1147 $Info{"${p}FileCntRep"} = 0;
1148 $Info{"${p}FileRepMax"} = 0;
1149 $Info{"${p}FileCntRename"} = 0;
1150 $Info{"${p}FileLinkMax"} = 0;
1151 $Info{"${p}Time"} = 0;
1152 for ( my $i = 0 ; $i < 16 ; $i++ ) {
1153 $Info{"${p}FileCnt"}
1154 += $Info{pool}{$p}[$i]{FileCnt};
1156 += $Info{pool}{$p}[$i]{DirCnt};
1158 += $Info{pool}{$p}[$i]{Kb};
1160 += $Info{pool}{$p}[$i]{Kb2};
1162 += $Info{pool}{$p}[$i]{KbRm};
1163 $Info{"${p}FileCntRm"}
1164 += $Info{pool}{$p}[$i]{FileCntRm};
1165 $Info{"${p}FileCntRep"}
1166 += $Info{pool}{$p}[$i]{FileCntRep};
1167 $Info{"${p}FileRepMax"}
1168 = $Info{pool}{$p}[$i]{FileRepMax}
1169 if ( $Info{"${p}FileRepMax"} <
1170 $Info{pool}{$p}[$i]{FileRepMax} );
1171 $Info{"${p}FileCntRename"}
1172 += $Info{pool}{$p}[$i]{FileCntRename};
1173 $Info{"${p}FileLinkMax"}
1174 = $Info{pool}{$p}[$i]{FileLinkMax}
1175 if ( $Info{"${p}FileLinkMax"} <
1176 $Info{pool}{$p}[$i]{FileLinkMax} );
1177 $Info{"${p}Time"} = $Info{pool}{$p}[$i]{Time}
1178 if ( $Info{"${p}Time"} <
1179 $Info{pool}{$p}[$i]{Time} );
1181 printf(LOG "%s%s nightly clean removed %d files of"
1183 $bpc->timeStamp, ucfirst($p),
1184 $Info{"${p}FileCntRm"},
1185 $Info{"${p}KbRm"} / (1000 * 1024));
1186 printf(LOG "%s%s is %.2fGB, %d files (%d repeated, "
1187 . "%d max chain, %d max links), %d directories\n",
1188 $bpc->timeStamp, ucfirst($p),
1189 $Info{"${p}Kb"} / (1000 * 1024),
1190 $Info{"${p}FileCnt"}, $Info{"${p}FileCntRep"},
1191 $Info{"${p}FileRepMax"},
1192 $Info{"${p}FileLinkMax"}, $Info{"${p}DirCnt"});
1200 # Queue BackupPC_link to complete the backup
1201 # processing for this host.
1203 if ( defined($Status{$host})
1204 && ($Status{$host}{reason} eq "Reason_backup_done"
1205 || $Status{$host}{needLink}) ) {
1207 } elsif ( defined($Status{$host}) ) {
1208 $Status{$host}{state} = "Status_idle";
1211 delete($Jobs{$host});
1212 $Status{$host}{activeJob} = 0 if ( defined($Status{$host}) );
1216 # When we are idle (empty Jobs, CmdQueue, BgQueue, UserQueue) we
1217 # do a pass over %Status updating the deadCnt and aliveCnt for
1218 # DHCP hosts. The reason we need to do this later is we can't
1219 # be sure whether a DHCP host is alive or dead until we have passed
1220 # over all the DHCP pool.
1222 return if ( @CmdQueue || @BgQueue || @UserQueue || keys(%Jobs) > 1 );
1223 foreach my $host ( keys(%Status) ) {
1224 next if ( $Status{$host}{dhcpCheckCnt} <= 0 );
1225 $Status{$host}{deadCnt} += $Status{$host}{dhcpCheckCnt};
1226 $Status{$host}{dhcpCheckCnt} = 0;
1227 if ( $Status{$host}{deadCnt} >= $Conf{BlackoutBadPingLimit} ) {
1228 $Status{$host}{aliveCnt} = 0;
1233 ############################################################################
1234 # Main_Check_Client_Messages($fdRead)
1236 # Check for, and process, any output from our clients. Also checks
1237 # for new connections to our SERVER_UNIX and SERVER_INET sockets.
1238 ############################################################################
1239 sub Main_Check_Client_Messages
1242 foreach my $client ( keys(%Clients) ) {
1243 next if ( !vec($fdRead, $Clients{$client}{fn}, 1) );
1246 # do a last check to make sure there is something to read so
1247 # we are absolutely sure we won't block.
1249 vec(my $readMask, $Clients{$client}{fn}, 1) = 1;
1250 if ( !select($readMask, undef, undef, 0.0) ) {
1251 print(LOG $bpc->timeStamp, "Botch in Main_Check_Client_Messages:"
1252 . " nothing to read from $client. Debug dump:\n");
1253 my($dump) = Data::Dumper->new(
1254 [ \%Clients, \%Jobs, \$FDread, \$fdRead],
1255 [qw(*Clients, *Jobs *FDread, *fdRead)]);
1257 print(LOG $dump->Dump);
1260 my $nbytes = sysread($Clients{$client}{fh}, $mesg, 1024);
1261 $Clients{$client}{mesg} .= $mesg if ( $nbytes > 0 );
1263 # Process any complete lines received from this client.
1265 while ( $Clients{$client}{mesg} =~ /(.*?)[\n\r]+(.*)/s ) {
1268 $Clients{$client}{mesg} = $2;
1270 # Authenticate the message by checking the MD5 digest
1272 my $md5 = Digest::MD5->new;
1273 if ( $cmd !~ /^(.{22}) (.*)/
1274 || ($md5->add($Clients{$client}{seed}
1275 . $Clients{$client}{mesgCnt}
1276 . $Conf{ServerMesgSecret} . $2),
1277 $md5->b64digest ne $1) ) {
1278 print(LOG $bpc->timeStamp, "Corrupted message '$cmd' from"
1279 . " client '$Clients{$client}{clientName}':"
1280 . " shutting down client connection\n");
1284 $Clients{$client}{mesgCnt}++;
1285 $cmd = decode_utf8($2);
1286 if ( $cmd =~ /^stop (\S+)\s+(\S+)\s+(\S*)/ ) {
1290 $host = $bpc->uriUnesc($host);
1291 if ( $CmdJob ne $host && defined($Status{$host})
1292 && defined($Jobs{$host}) ) {
1293 print(LOG $bpc->timeStamp,
1294 "Stopping current $Jobs{$host}{type} of $host,"
1295 . " request by $user (backoff=$backoff)\n");
1296 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
1298 # Don't close the pipe now; wait until the child
1299 # really exits later. Otherwise close() will
1300 # block until the child has exited.
1302 ##vec($FDread, $Jobs{$host}{fn}, 1) = 0;
1303 ##close($Jobs{$host}{fh});
1304 ##delete($Jobs{$host});
1306 $Status{$host}{state} = "Status_idle";
1307 if ( $Jobs{$host}{type} eq "restore" ) {
1308 $Status{$host}{reason}
1309 = "Reason_restore_canceled_by_user";
1310 } elsif ( $Jobs{$host}{type} eq "archive" ) {
1311 $Status{$host}{reason}
1312 = "Reason_archive_canceled_by_user";
1314 $Status{$host}{reason}
1315 = "Reason_backup_canceled_by_user";
1317 $Status{$host}{activeJob} = 0;
1318 $Status{$host}{startTime} = time;
1319 $reply = "ok: $Jobs{$host}{type} of $host canceled";
1320 } elsif ( $BgQueueOn{$host} || $UserQueueOn{$host} ) {
1321 print(LOG $bpc->timeStamp,
1322 "Stopping pending backup of $host,"
1323 . " request by $user (backoff=$backoff)\n");
1324 @BgQueue = grep($_->{host} ne $host, @BgQueue);
1325 @UserQueue = grep($_->{host} ne $host, @UserQueue);
1326 $BgQueueOn{$host} = $UserQueueOn{$host} = 0;
1327 $reply = "ok: pending backup of $host canceled";
1329 print(LOG $bpc->timeStamp,
1330 "Nothing to do for stop backup of $host,"
1331 . " request by $user (backoff=$backoff)\n");
1332 $reply = "ok: no backup was pending or running";
1334 if ( defined($Status{$host}) && $backoff ne "" ) {
1335 if ( $backoff > 0 ) {
1336 $Status{$host}{backoffTime} = time + $backoff * 3600;
1338 delete($Status{$host}{backoffTime});
1341 } elsif ( $cmd =~ /^backup all$/ ) {
1343 } elsif ( $cmd =~ /^BackupPC_nightly run$/ ) {
1344 $RunNightlyWhenIdle = 1;
1345 } elsif ( $cmd =~ /^queue (\S+)$/ ) {
1347 $host = $bpc->uriUnesc($host);
1348 if ( !defined($Hosts->{$host}) ) {
1349 print(LOG $bpc->timeStamp,
1350 "User requested backup of unknown host $host\n");
1351 $reply = "error: unknown host $host";
1353 if ( QueueOnePC($host) ) {
1354 print(LOG $bpc->timeStamp,
1355 "Disk too full ($Info{DUlastValue}%); skipped 1 host\n");
1356 $Info{DUDailySkipHostCnt}++;
1357 $reply = "error: disk too full to queue $host";
1359 print(LOG $bpc->timeStamp, "Host $host queued by user.\n");
1360 $reply = "ok: $host queued";
1363 } elsif ( $cmd =~ /^backup (\S+)\s+(\S+)\s+(\S+)\s+(\S+)/ ) {
1368 $host = $bpc->uriUnesc($host);
1369 $hostIP = $bpc->uriUnesc($hostIP);
1370 if ( !defined($Hosts->{$host}) ) {
1371 print(LOG $bpc->timeStamp,
1372 "User $user requested backup of unknown host"
1374 $reply = "error: unknown host $host";
1376 print(LOG $bpc->timeStamp,
1377 "User $user requested backup of $host"
1379 if ( $BgQueueOn{$hostIP} ) {
1380 @BgQueue = grep($_->{host} ne $hostIP, @BgQueue);
1381 $BgQueueOn{$hostIP} = 0;
1383 if ( $UserQueueOn{$hostIP} ) {
1384 @UserQueue = grep($_->{host} ne $hostIP, @UserQueue);
1385 $UserQueueOn{$hostIP} = 0;
1387 unshift(@UserQueue, {
1393 dhcp => $hostIP eq $host ? 0 : 1,
1395 $UserQueueOn{$hostIP} = 1;
1396 $reply = "ok: requested backup of $host";
1398 } elsif ( $cmd =~ /^archive (\S+)\s+(\S+)\s+(\S+)/ ) {
1400 my $archivehost = $2;
1401 my $reqFileName = $3;
1402 $host = $bpc->uriUnesc($archivehost);
1403 if ( !defined($Status{$host}) ) {
1404 print(LOG $bpc->timeStamp,
1405 "User $user requested archive of unknown archive host"
1407 $reply = "archive error: unknown archive host $host";
1409 print(LOG $bpc->timeStamp,
1410 "User $user requested archive on $host"
1412 if ( defined($Jobs{$host}) ) {
1413 $reply = "Archive currently running on $host, please try later";
1415 unshift(@UserQueue, {
1418 reqFileName => $reqFileName,
1424 $UserQueueOn{$host} = 1;
1425 $reply = "ok: requested archive on $host";
1428 } elsif ( $cmd =~ /^restore (\S+)\s+(\S+)\s+(\S+)\s+(\S+)/ ) {
1432 my $reqFileName = $4;
1433 $host = $bpc->uriUnesc($host);
1434 $hostIP = $bpc->uriUnesc($hostIP);
1435 if ( !defined($Hosts->{$host}) ) {
1436 print(LOG $bpc->timeStamp,
1437 "User $user requested restore to unknown host"
1439 $reply = "restore error: unknown host $host";
1441 print(LOG $bpc->timeStamp,
1442 "User $user requested restore to $host"
1444 unshift(@UserQueue, {
1447 reqFileName => $reqFileName,
1453 $UserQueueOn{$host} = 1;
1454 if ( defined($Jobs{$host}) ) {
1455 $reply = "ok: requested restore of $host, but a"
1456 . " job is currently running,"
1457 . " so this request will start later";
1459 $reply = "ok: requested restore of $host";
1462 } elsif ( $cmd =~ /^status\s*(.*)/ ) {
1464 my($dump, @values, @names);
1465 foreach my $type ( split(/\s+/, $args) ) {
1466 if ( $type =~ /^queues/ ) {
1467 push(@values, \@BgQueue, \@UserQueue, \@CmdQueue);
1468 push(@names, qw(*BgQueue *UserQueue *CmdQueue));
1469 } elsif ( $type =~ /^jobs/ ) {
1470 push(@values, \%Jobs);
1471 push(@names, qw(*Jobs));
1472 } elsif ( $type =~ /^queueLen/ ) {
1474 BgQueue => scalar(@BgQueue),
1475 UserQueue => scalar(@UserQueue),
1476 CmdQueue => scalar(@CmdQueue),
1478 push(@names, qw(*QueueLen));
1479 } elsif ( $type =~ /^info/ ) {
1480 push(@values, \%Info);
1481 push(@names, qw(*Info));
1482 } elsif ( $type =~ /^hosts/ ) {
1483 push(@values, \%Status);
1484 push(@names, qw(*Status));
1485 } elsif ( $type =~ /^host\((.*)\)/ ) {
1486 my $h = $bpc->uriUnesc($1);
1487 if ( defined($Status{$h}) ) {
1490 BgQueueOn => $BgQueueOn{$h},
1491 UserQueueOn => $UserQueueOn{$h},
1492 CmdQueueOn => $CmdQueueOn{$h},
1494 push(@names, qw(*StatusHost));
1496 print(LOG $bpc->timeStamp,
1497 "Unknown host $h for status request\n");
1500 print(LOG $bpc->timeStamp,
1501 "Unknown status request $type\n");
1504 $dump = Data::Dumper->new(\@values, \@names);
1506 $reply = $dump->Dump;
1507 } elsif ( $cmd =~ /^link\s+(.+)/ ) {
1509 $host = $bpc->uriUnesc($host);
1511 } elsif ( $cmd =~ /^log\s+(.*)/ ) {
1512 print(LOG $bpc->timeStamp, "$1\n");
1513 } elsif ( $cmd =~ /^server\s+(\w+)/ ) {
1515 if ( $type eq 'reload' ) {
1516 ServerReload("Reloading config/host files via CGI request");
1517 } elsif ( $type eq 'shutdown' ) {
1518 $reply = "Shutting down...\n";
1519 syswrite($Clients{$client}{fh}, $reply, length($reply));
1520 ServerShutdown("Server shutting down...");
1522 } elsif ( $cmd =~ /^quit/ || $cmd =~ /^exit/ ) {
1526 print(LOG $bpc->timeStamp, "Unknown command $cmd\n");
1527 $reply = "error: bad command $cmd";
1530 # send a reply to the client, at a minimum "ok\n".
1532 $reply = "ok" if ( $reply eq "" );
1534 syswrite($Clients{$client}{fh}, $reply, length($reply));
1537 # Detect possible denial-of-service attack from sending a huge line
1538 # (ie: never terminated). 32K seems to be plenty big enough as
1541 if ( length($Clients{$client}{mesg}) > 32 * 1024 ) {
1542 print(LOG $bpc->timeStamp, "Line too long from client"
1543 . " '$Clients{$client}{clientName}':"
1544 . " shutting down client connection\n");
1548 # Shut down the client connection if we read EOF
1550 if ( $nbytes <= 0 ) {
1551 close($Clients{$client}{fh});
1552 vec($FDread, $Clients{$client}{fn}, 1) = 0;
1553 delete($Clients{$client});
1557 # Accept any new connections on each of our listen sockets
1559 if ( vec($fdRead, fileno(SERVER_UNIX), 1) ) {
1561 my $paddr = accept(CLIENT, SERVER_UNIX);
1563 $Clients{$ClientConnCnt}{clientName} = "unix socket";
1564 $Clients{$ClientConnCnt}{mesg} = "";
1565 $Clients{$ClientConnCnt}{fh} = *CLIENT;
1566 $Clients{$ClientConnCnt}{fn} = fileno(CLIENT);
1567 vec($FDread, $Clients{$ClientConnCnt}{fn}, 1) = 1;
1569 # Generate and send unique seed for MD5 digests to avoid
1570 # replay attacks. See BackupPC::Lib::ServerMesg().
1572 my $seed = time . ",$ClientConnCnt,$$,0\n";
1573 $Clients{$ClientConnCnt}{seed} = $seed;
1574 $Clients{$ClientConnCnt}{mesgCnt} = 0;
1575 syswrite($Clients{$ClientConnCnt}{fh}, $seed, length($seed));
1577 if ( $ServerInetPort > 0 && vec($fdRead, fileno(SERVER_INET), 1) ) {
1579 my $paddr = accept(CLIENT, SERVER_INET);
1580 my($port,$iaddr) = sockaddr_in($paddr);
1581 my $name = gethostbyaddr($iaddr, AF_INET);
1583 $Clients{$ClientConnCnt}{mesg} = "";
1584 $Clients{$ClientConnCnt}{fh} = *CLIENT;
1585 $Clients{$ClientConnCnt}{fn} = fileno(CLIENT);
1586 $Clients{$ClientConnCnt}{clientName} = "$name:$port";
1587 vec($FDread, $Clients{$ClientConnCnt}{fn}, 1) = 1;
1589 # Generate and send unique seed for MD5 digests to avoid
1590 # replay attacks. See BackupPC::Lib::ServerMesg().
1592 my $seed = time . ",$ClientConnCnt,$$,$port\n";
1593 $Clients{$ClientConnCnt}{seed} = $seed;
1594 $Clients{$ClientConnCnt}{mesgCnt} = 0;
1595 syswrite($Clients{$ClientConnCnt}{fh}, $seed, length($seed));
1599 ###########################################################################
1600 # Miscellaneous subroutines
1601 ###########################################################################
1604 # Write the current status to $LogDir/status.pl
1608 my($dump) = Data::Dumper->new(
1609 [ \%Info, \%Status],
1610 [qw(*Info *Status)]);
1612 my $text = $dump->Dump;
1613 $bpc->{storage}->TextFileWrite("$LogDir/status.pl", $text);
1617 # Compare function for host sort. Hosts with errors go first,
1618 # sorted with the oldest errors first. The remaining hosts
1619 # are sorted so that those with the oldest backups go first.
1624 # Hosts with errors go before hosts without errors
1626 return -1 if ( $Status{$a}{error} ne "" && $Status{$b}{error} eq "" );
1629 # Hosts with no errors go after hosts with errors
1631 return 1 if ( $Status{$a}{error} eq "" && $Status{$b}{error} ne "" );
1634 # hosts with the older last good backups sort earlier
1636 my $r = $Status{$a}{lastGoodBackupTime} <=> $Status{$b}{lastGoodBackupTime};
1637 return $r if ( $r );
1640 # Finally, just sort based on host name
1650 delete($Status{$host}{backoffTime})
1651 if ( defined($Status{$host}{backoffTime})
1652 && $Status{$host}{backoffTime} < time );
1653 return if ( defined($Jobs{$host})
1654 || $BgQueueOn{$host}
1655 || $UserQueueOn{$host}
1656 || $CmdQueueOn{$host} );
1657 if ( $Hosts->{$host}{dhcp} ) {
1658 $Status{$host}{dhcpCheckCnt}++;
1659 if ( $RunNightlyWhenIdle ) {
1661 # Once per night queue a check for DHCP hosts that just
1662 # checks for expired dumps. We need to do this to handle
1663 # the case when a DHCP host has not been on the network for
1664 # a long time, and some of the old dumps need to be expired.
1665 # Normally expiry checks are done by BackupPC_dump only
1666 # after the DHCP hosts has been detected on the network.
1669 {host => $host, user => "BackupPC", reqTime => time,
1670 dhcp => 0, dumpExpire => 1});
1671 $BgQueueOn{$host} = 1;
1675 # this is a fixed ip host: queue it
1677 if ( $Info{DUlastValue} > $Conf{DfMaxUsagePct} ) {
1679 # Since we are out of disk space, instead of queuing
1680 # a regular job, queue an expire check instead. That
1681 # way if the admin reduces the number of backups to
1682 # keep then we will actually delete them. Otherwise
1683 # BackupPC_dump will never run since we have exceeded
1688 {host => $host, user => "BackupPC", reqTime => time,
1689 dhcp => $Hosts->{$host}{dhcp}, dumpExpire => 1});
1692 # Queue regular background backup
1695 {host => $host, user => "BackupPC", reqTime => time,
1696 dhcp => $Hosts->{$host}{dhcp}});
1698 $BgQueueOn{$host} = 1;
1705 # Queue all the hosts for backup. This means queuing all the fixed
1706 # ip hosts and all the dhcp address ranges. We also additionally
1707 # queue the dhcp hosts with a -e flag to check for expired dumps.
1713 foreach my $host ( sort(HostSortCompare keys(%$Hosts)) ) {
1714 $nSkip += QueueOnePC($host);
1717 print(LOG $bpc->timeStamp,
1718 "Disk too full ($Info{DUlastValue}%); skipped $nSkip hosts\n");
1719 $Info{DUDailySkipHostCnt} += $nSkip;
1721 foreach my $dhcp ( @{$Conf{DHCPAddressRanges}} ) {
1722 for ( my $i = $dhcp->{first} ; $i <= $dhcp->{last} ; $i++ ) {
1723 my $ipAddr = "$dhcp->{ipAddrBase}.$i";
1724 next if ( defined($Jobs{$ipAddr})
1725 || $BgQueueOn{$ipAddr}
1726 || $UserQueueOn{$ipAddr}
1727 || $CmdQueueOn{$ipAddr} );
1729 # this is a potential dhcp ip address (we don't know the
1730 # host name yet): queue it
1733 {host => $ipAddr, user => "BackupPC", reqTime => time,
1735 $BgQueueOn{$ipAddr} = 1;
1741 # Queue a BackupPC_link for the given host
1747 return if ( $CmdQueueOn{$host} );
1748 $Status{$host}{state} = "Status_link_pending";
1749 $Status{$host}{needLink} = 0;
1750 unshift(@CmdQueue, {
1754 cmd => ["$BinDir/BackupPC_link", $host],
1756 $CmdQueueOn{$host} = 1;
1760 # Read the hosts file, and update Status if any hosts have been
1761 # added or deleted. We also track the mtime so the only need to
1762 # update the hosts file on changes.
1764 # This function is called at startup, SIGHUP, and on each wakeup.
1765 # It returns 1 on success and undef on failure.
1772 # Nothing to do if we already have the current hosts file
1774 return 1 if ( !$force && defined($Hosts)
1775 && $Info{HostsModTime} == $bpc->HostsMTime() );
1776 if ( !defined($newHosts = $bpc->HostInfoRead()) ) {
1777 print(LOG $bpc->timeStamp, "Can't read hosts file!\n");
1780 print(LOG $bpc->timeStamp, "Reading hosts file\n");
1782 $Info{HostsModTime} = $bpc->HostsMTime();
1784 # Now update %Status in case any hosts have been added or deleted
1786 foreach my $host ( sort(keys(%$Hosts)) ) {
1787 next if ( defined($Status{$host}) );
1788 $Status{$host}{state} = "Status_idle";
1789 print(LOG $bpc->timeStamp, "Added host $host to backup list\n");
1791 foreach my $host ( sort(keys(%Status)) ) {
1792 next if ( $host eq $bpc->trashJob
1793 || $bpc->isAdminJob($host)
1794 || defined($Hosts->{$host})
1795 || defined($Jobs{$host})
1796 || $BgQueueOn{$host}
1797 || $UserQueueOn{$host}
1798 || $CmdQueueOn{$host} );
1799 print(LOG $bpc->timeStamp, "Deleted host $host from backup list\n");
1800 delete($Status{$host});
1806 # Remember the signal name for later processing
1812 foreach my $host ( keys(%Jobs) ) {
1813 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
1816 # In case we are inside the exit handler, reopen the log file
1820 print(LOG "Fatal error: unhandled signal $SigName\n");
1821 unlink("$LogDir/BackupPC.pid");
1822 confess("Got new signal $SigName... quitting\n");
1829 # Open the log file and point STDOUT and STDERR there too
1833 mkpath($LogDir, 0, 0777) if ( !-d $LogDir );
1834 open(LOG, ">>$LogDir/LOG")
1835 || die("Can't create LOG file $LogDir/LOG");
1838 open(STDOUT, ">&LOG");
1839 open(STDERR, ">&LOG");
1840 select(LOG); $| = 1;
1841 select(STDERR); $| = 1;
1842 select(STDOUT); $| = 1;
1846 # Initialize the unix-domain and internet-domain sockets that
1847 # we listen to for client connections (from the CGI script and
1848 # some of the BackupPC sub-programs).
1850 sub ServerSocketInit
1852 if ( !defined(fileno(SERVER_UNIX)) ) {
1854 # one-time only: initialize unix-domain socket
1856 if ( !socket(SERVER_UNIX, PF_UNIX, SOCK_STREAM, 0) ) {
1857 print(LOG $bpc->timeStamp, "unix socket() failed: $!\n");
1860 my $sockFile = "$LogDir/BackupPC.sock";
1862 if ( !bind(SERVER_UNIX, sockaddr_un($sockFile)) ) {
1863 print(LOG $bpc->timeStamp, "unix bind() failed: $!\n");
1866 if ( !listen(SERVER_UNIX, SOMAXCONN) ) {
1867 print(LOG $bpc->timeStamp, "unix listen() failed: $!\n");
1870 vec($FDread, fileno(SERVER_UNIX), 1) = 1;
1872 return if ( $ServerInetPort == $Conf{ServerPort} );
1873 if ( $ServerInetPort > 0 ) {
1874 vec($FDread, fileno(SERVER_INET), 1) = 0;
1876 $ServerInetPort = -1;
1878 if ( $Conf{ServerPort} > 0 ) {
1880 # Setup a socket to listen on $Conf{ServerPort}
1882 my $proto = getprotobyname('tcp');
1883 if ( !socket(SERVER_INET, PF_INET, SOCK_STREAM, $proto) ) {
1884 print(LOG $bpc->timeStamp, "inet socket() failed: $!\n");
1887 if ( !setsockopt(SERVER_INET, SOL_SOCKET, SO_REUSEADDR, pack("l",1)) ) {
1888 print(LOG $bpc->timeStamp, "setsockopt() failed: $!\n");
1891 if ( !bind(SERVER_INET, sockaddr_in($Conf{ServerPort}, INADDR_ANY)) ) {
1892 print(LOG $bpc->timeStamp, "inet bind() failed: $!\n");
1895 if ( !listen(SERVER_INET, SOMAXCONN) ) {
1896 print(LOG $bpc->timeStamp, "inet listen() failed: $!\n");
1899 vec($FDread, fileno(SERVER_INET), 1) = 1;
1900 $ServerInetPort = $Conf{ServerPort};
1905 # Reload the server. Used by Main_Process_Signal when $SigName eq "HUP"
1906 # or when the command "server reload" is received.
1911 $mesg = $bpc->ConfigRead() || $mesg;
1912 print(LOG $bpc->timeStamp, "$mesg\n");
1913 $Info{ConfigModTime} = $bpc->ConfigMTime();
1914 %Conf = $bpc->Conf();
1915 umask($Conf{UmaskMode});
1919 $Info{ConfigLTime} = time;
1923 # Gracefully shutdown the server. Used by Main_Process_Signal when
1924 # $SigName ne "" && $SigName ne "HUP" or when the command
1925 # "server shutdown" is received.
1930 print(LOG $bpc->timeStamp, "$mesg\n");
1931 if ( keys(%Jobs) ) {
1932 foreach my $host ( keys(%Jobs) ) {
1933 kill($bpc->sigName2num("INT"), $Jobs{$host}{pid});
1936 foreach my $host ( keys(%Jobs) ) {
1937 kill($bpc->sigName2num("KILL"), $Jobs{$host}{pid});
1943 unlink("$LogDir/BackupPC.pid");