X-Git-Url: http://git.rot13.org/?p=BackupPC.git;a=blobdiff_plain;f=bin%2FBackupPC;h=79d89c74556c2f4b7a182962e6d1e1c1f1896ad4;hp=dab51db8b2a4f5ec01d8ccf55a4ad47b59562329;hb=2441b9094f3f6e4f2a3a4fe67781780e6f6890bd;hpb=d13d57e035dac9362ca393991b978530402969b7 diff --git a/bin/BackupPC b/bin/BackupPC index dab51db..79d89c7 100755 --- a/bin/BackupPC +++ b/bin/BackupPC @@ -6,7 +6,7 @@ # DESCRIPTION # # BackupPC reads the configuration and status information from -# $TopDir/conf. It then runs and manages all the backup activity. +# $ConfDir/conf. It then runs and manages all the backup activity. # # As specified by $Conf{WakeupSchedule}, BackupPC wakes up periodically # to queue backups on all the PCs. This is a four step process: @@ -29,7 +29,7 @@ # Craig Barratt # # COPYRIGHT -# Copyright (C) 2001-2003 Craig Barratt +# Copyright (C) 2001-2007 Craig Barratt # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -47,7 +47,7 @@ # #======================================================================== # -# Version 2.1.0, released 20 Jun 2004. +# Version 3.2.0beta0, released 17 Jan 2009. # # See http://backuppc.sourceforge.net. # @@ -59,6 +59,7 @@ use vars qw(%Status %Info $Hosts); use lib "/usr/local/BackupPC/lib"; use BackupPC::Lib; use BackupPC::FileZIO; +use Encode qw/decode_utf8/; use File::Path; use Data::Dumper; @@ -66,6 +67,7 @@ use Getopt::Std; use Socket; use Carp; use Digest::MD5; +use POSIX qw(setsid); ########################################################################### # Handle command line options @@ -86,6 +88,7 @@ if ( !getopts("d", \%opts) || @ARGV != 0 ) { die("BackupPC::Lib->new failed\n") if ( !(my $bpc = BackupPC::Lib->new) ); my $TopDir = $bpc->TopDir(); my $BinDir = $bpc->BinDir(); +my $LogDir = $bpc->LogDir(); my %Conf = $bpc->Conf(); # @@ -110,10 +113,14 @@ if ( $Conf{BackupPCUserVerify} # # Read old status # -if ( -f "$TopDir/log/status.pl" && !(my $ret = do "$TopDir/log/status.pl") ) { - die "couldn't parse $TopDir/log/status.pl: $@" if $@; - die "couldn't do $TopDir/log/status.pl: $!" unless defined $ret; - die "couldn't run $TopDir/log/status.pl"; +if ( -f "$LogDir/status.pl" && !(my $ret = do "$LogDir/status.pl") ) { + if ( $@ ) { + print STDERR "couldn't parse $LogDir/status.pl: $@"; + } elsif ( !defined($ret) ) { + print STDERR "couldn't do $LogDir/status.pl: $!"; + } else { + print STDERR "couldn't run $LogDir/status.pl"; + } } # @@ -241,14 +248,16 @@ sub Main_Initialize umask($Conf{UmaskMode}); # - # Check for another running process, check that PASSWD is set and - # verify executables are configured correctly. + # Check for another running process, verify executables are configured + # correctly and make sure $TopDir is on a file system that supports + # hardlinks. # if ( $Info{pid} ne "" && kill(0, $Info{pid}) ) { print(STDERR $bpc->timeStamp, "Another BackupPC is running (pid $Info{pid}); quitting...\n"); exit(1); } + foreach my $progName ( qw(SmbClientPath NmbLookupPath PingPath DfPath SendmailPath SshPath) ) { next if ( $Conf{$progName} eq "" || -x $Conf{$progName} ); @@ -258,12 +267,35 @@ sub Main_Initialize exit(1); } + if ( !$bpc->HardlinkTest("$TopDir/pc", "$TopDir/cpool") ) { + print(STDERR $bpc->timeStamp, "Can't create a test hardlink between a file" + . " in $TopDir/pc and $TopDir/cpool. Either these are different" + . " file systems, or this file system doesn't support hardlinks," + . " or these directories don't exist, or there is a permissions" + . " problem, or the file system is out of inodes or full. Use" + . " df, df -i, and ls -ld to check each of these possibilities." + . " Quitting...\n"); + exit(1); + } + if ( $opts{d} ) { # - # daemonize by forking + # daemonize by forking; more robust method per: + # http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=301057 # - defined(my $pid = fork) or die "Can't fork: $!"; + my $pid; + defined($pid = fork) or die("Can't fork: $!"); + exit if( $pid ); # parent exits + + POSIX::setsid(); + defined($pid = fork) or die("Can't fork: $!"); exit if $pid; # parent exits + + chdir ("/") or die("Cannot chdir to /: $!\n"); + close(STDIN); + open(STDIN , ">/dev/null") or die("Cannot open /dev/null as stdin\n"); + # STDOUT and STDERR are handled in LogFileOpen() right below, + # otherwise we would have to reopen them too. } # @@ -331,9 +363,11 @@ sub Main_Initialize # Write out our initial status and save our PID # StatusWrite(); - if ( open(PID, ">", "$TopDir/log/BackupPC.pid") ) { + unlink("$LogDir/BackupPC.pid"); + if ( open(PID, ">", "$LogDir/BackupPC.pid") ) { print(PID $$); close(PID); + chmod(0444, "$LogDir/BackupPC.pid"); } # @@ -374,8 +408,7 @@ sub Main_TryToRun_nightly }); $CmdQueueOn{$bpc->trashJob} = 1; } - if ( keys(%Jobs) == $trashCleanRunning && $RunNightlyWhenIdle == 1 ) { - + if ( $RunNightlyWhenIdle == 1 ) { # # Queue multiple nightly jobs based on the configuration # @@ -403,9 +436,6 @@ sub Main_TryToRun_nightly # # Zero out the data we expect to get from BackupPC_nightly. - # In the future if we want to split BackupPC_nightly over - # more than one night we will only zero out the portion - # that we are running right now. # for my $p ( qw(pool cpool) ) { for ( my $i = $start ; $i < $end ; $i++ ) { @@ -429,7 +459,7 @@ sub Main_TryToRun_nightly # # Now queue the $Conf{MaxBackupPCNightlyJobs} jobs. - # The granularity on start and end is now 0..256. + # The granularity on start and end is now 0..255. # $start *= 16; $end *= 16; @@ -446,7 +476,6 @@ sub Main_TryToRun_nightly $start = $start0 + int(($end - $start0) * ($i + 1) / $Conf{MaxBackupPCNightlyJobs}); push(@$cmd, $start - 1); - my $job = $bpc->adminJob($i); unshift(@CmdQueue, { host => $job, @@ -457,7 +486,6 @@ sub Main_TryToRun_nightly $CmdQueueOn{$job} = 1; } $RunNightlyWhenIdle = 2; - } } @@ -505,6 +533,8 @@ sub Main_TryToRun_CmdQueue } if ( !$pid ) { setpgrp 0,0; + $ENV{BPC_REQUSER} = $req->{user}; + POSIX::nice($Conf{CmdQueueNice}) if ( $Conf{CmdQueueNice} ); exec(@$cmd); print(LOG $bpc->timeStamp, "can't exec @$cmd for $host\n"); exit(0); @@ -596,7 +626,14 @@ sub Main_TryToRun_Bg_or_User_Queue } } - while ( $RunNightlyWhenIdle == 0 ) { + # + # Run background jobs anytime. Previously they were locked out + # when BackupPC_nightly was running or pending with this + # condition on the while loop: + # + # while ( $RunNightlyWhenIdle == 0 ) + # + while ( 1 ) { local(*FH); my(@args, $progName, $type); my $nJobs = keys(%Jobs); @@ -794,15 +831,15 @@ sub Main_Check_Timeout $Info{DUDailySkipHostCntPrev} = $Info{DUDailySkipHostCnt}; $Info{DUDailySkipHostCnt} = 0; my $lastLog = $Conf{MaxOldLogFiles} - 1; - if ( -f "$TopDir/log/LOG.$lastLog" ) { + if ( -f "$LogDir/LOG.$lastLog" ) { print(LOG $bpc->timeStamp, - "Removing $TopDir/log/LOG.$lastLog\n"); - unlink("$TopDir/log/LOG.$lastLog"); + "Removing $LogDir/LOG.$lastLog\n"); + unlink("$LogDir/LOG.$lastLog"); } - if ( -f "$TopDir/log/LOG.$lastLog.z" ) { + if ( -f "$LogDir/LOG.$lastLog.z" ) { print(LOG $bpc->timeStamp, - "Removing $TopDir/log/LOG.$lastLog.z\n"); - unlink("$TopDir/log/LOG.$lastLog.z"); + "Removing $LogDir/LOG.$lastLog.z\n"); + unlink("$LogDir/LOG.$lastLog.z"); } print(LOG $bpc->timeStamp, "Aging LOG files, LOG -> LOG.0 -> " . "LOG.1 -> ... -> LOG.$lastLog\n"); @@ -811,24 +848,30 @@ sub Main_Check_Timeout close(LOG); for ( my $i = $lastLog - 1 ; $i >= 0 ; $i-- ) { my $j = $i + 1; - rename("$TopDir/log/LOG.$i", "$TopDir/log/LOG.$j") - if ( -f "$TopDir/log/LOG.$i" ); - rename("$TopDir/log/LOG.$i.z", "$TopDir/log/LOG.$j.z") - if ( -f "$TopDir/log/LOG.$i.z" ); + rename("$LogDir/LOG.$i", "$LogDir/LOG.$j") + if ( -f "$LogDir/LOG.$i" ); + rename("$LogDir/LOG.$i.z", "$LogDir/LOG.$j.z") + if ( -f "$LogDir/LOG.$i.z" ); } # # Compress the log file LOG -> LOG.0.z (if enabled). # Otherwise, just rename LOG -> LOG.0. # - BackupPC::FileZIO->compressCopy("$TopDir/log/LOG", - "$TopDir/log/LOG.0.z", - "$TopDir/log/LOG.0", + BackupPC::FileZIO->compressCopy("$LogDir/LOG", + "$LogDir/LOG.0.z", + "$LogDir/LOG.0", $Conf{CompressLevel}, 1); LogFileOpen(); # - # Remember to run nightly script after current jobs are done + # Remember to run the nightly script when the next CmdQueue + # job is done. # - $RunNightlyWhenIdle = 1; + if ( $RunNightlyWhenIdle == 2 ) { + print(LOG $bpc->timeStamp, "BackupPC_nightly is still running after 24 hours!!" + . " You should adjust the config settings; Skipping this run\n"); + } else { + $RunNightlyWhenIdle = 1; + } } # # Write out the current status and then queue all the PCs @@ -918,6 +961,8 @@ sub Main_Check_Job_Messages if ( $Status{$host}{dhcpCheckCnt} > 0 ); } elsif ( $mesg =~ /^xferPids (.*)/ ) { $Jobs{$host}{xferPid} = $1; + } elsif ( $mesg =~ /^completionPercent (.*)/ ) { + $Jobs{$host}{completionPercent} = $1; } elsif ( $mesg =~ /^started_restore/ ) { $Jobs{$host}{type} = "restore"; print(LOG $bpc->timeStamp, @@ -1057,7 +1102,8 @@ sub Main_Check_Job_Messages # # This means the last BackupPC_nightly is done with # the pool clean, so it's ok to start running regular - # backups again. + # backups again. But starting in 3.0 regular jobs + # are decoupled from BackupPC_nightly. # $RunNightlyWhenIdle = 0; } @@ -1087,6 +1133,9 @@ sub Main_Check_Job_Messages #print(LOG $bpc->timeStamp, "BackupPC_nightly done; now" # . " have $BackupPCNightlyJobs running\n"); if ( $BackupPCNightlyJobs <= 0 ) { + # + # Last BackupPC_nightly has finished + # $BackupPCNightlyJobs = 0; $RunNightlyWhenIdle = 0; $CmdJob = ""; @@ -1238,7 +1287,7 @@ sub Main_Check_Client_Messages last; } $Clients{$client}{mesgCnt}++; - $cmd = $2; + $cmd = decode_utf8($2); if ( $cmd =~ /^stop (\S+)\s+(\S+)\s+(\S*)/ ) { $host = $1; my $user = $2; @@ -1298,6 +1347,24 @@ sub Main_Check_Client_Messages QueueAllPCs(); } elsif ( $cmd =~ /^BackupPC_nightly run$/ ) { $RunNightlyWhenIdle = 1; + } elsif ( $cmd =~ /^queue (\S+)$/ ) { + $host = $1; + $host = $bpc->uriUnesc($host); + if ( !defined($Hosts->{$host}) ) { + print(LOG $bpc->timeStamp, + "User requested backup of unknown host $host\n"); + $reply = "error: unknown host $host"; + } else { + if ( QueueOnePC($host) ) { + print(LOG $bpc->timeStamp, + "Disk too full ($Info{DUlastValue}%); skipped 1 host\n"); + $Info{DUDailySkipHostCnt}++; + $reply = "error: disk too full to queue $host"; + } else { + print(LOG $bpc->timeStamp, "Host $host queued by user.\n"); + $reply = "ok: $host queued"; + } + } } elsif ( $cmd =~ /^backup (\S+)\s+(\S+)\s+(\S+)\s+(\S+)/ ) { my $hostIP = $1; $host = $2; @@ -1305,7 +1372,7 @@ sub Main_Check_Client_Messages my $doFull = $4; $host = $bpc->uriUnesc($host); $hostIP = $bpc->uriUnesc($hostIP); - if ( !defined($Status{$host}) ) { + if ( !defined($Hosts->{$host}) ) { print(LOG $bpc->timeStamp, "User $user requested backup of unknown host" . " $host\n"); @@ -1370,7 +1437,7 @@ sub Main_Check_Client_Messages my $reqFileName = $4; $host = $bpc->uriUnesc($host); $hostIP = $bpc->uriUnesc($hostIP); - if ( !defined($Status{$host}) ) { + if ( !defined($Hosts->{$host}) ) { print(LOG $bpc->timeStamp, "User $user requested restore to unknown host" . " $host"); @@ -1539,7 +1606,7 @@ sub Main_Check_Client_Messages ########################################################################### # -# Write the current status to $TopDir/log/status.pl +# Write the current status to $LogDir/status.pl # sub StatusWrite { @@ -1547,10 +1614,8 @@ sub StatusWrite [ \%Info, \%Status], [qw(*Info *Status)]); $dump->Indent(1); - if ( open(STATUS, ">", "$TopDir/log/status.pl") ) { - print(STATUS $dump->Dump); - close(STATUS); - } + my $text = $dump->Dump; + $bpc->{storage}->TextFileWrite("$LogDir/status.pl", $text); } # @@ -1568,7 +1633,6 @@ sub HostSortCompare # # Hosts with no errors go after hosts with errors # - return 1 if ( $Status{$a}{error} eq "" && $Status{$b}{error} ne "" ); # @@ -1583,6 +1647,65 @@ sub HostSortCompare return $a cmp $b; } +sub QueueOnePC +{ + my($host) = @_; + my $skipped = 0; + + delete($Status{$host}{backoffTime}) + if ( defined($Status{$host}{backoffTime}) + && $Status{$host}{backoffTime} < time ); + return if ( defined($Jobs{$host}) + || $BgQueueOn{$host} + || $UserQueueOn{$host} + || $CmdQueueOn{$host} ); + if ( $Hosts->{$host}{dhcp} ) { + $Status{$host}{dhcpCheckCnt}++; + if ( $RunNightlyWhenIdle ) { + # + # Once per night queue a check for DHCP hosts that just + # checks for expired dumps. We need to do this to handle + # the case when a DHCP host has not been on the network for + # a long time, and some of the old dumps need to be expired. + # Normally expiry checks are done by BackupPC_dump only + # after the DHCP hosts has been detected on the network. + # + unshift(@BgQueue, + {host => $host, user => "BackupPC", reqTime => time, + dhcp => 0, dumpExpire => 1}); + $BgQueueOn{$host} = 1; + } + } else { + # + # this is a fixed ip host: queue it + # + if ( $Info{DUlastValue} > $Conf{DfMaxUsagePct} ) { + # + # Since we are out of disk space, instead of queuing + # a regular job, queue an expire check instead. That + # way if the admin reduces the number of backups to + # keep then we will actually delete them. Otherwise + # BackupPC_dump will never run since we have exceeded + # the limit. + # + $skipped = 1; + unshift(@BgQueue, + {host => $host, user => "BackupPC", reqTime => time, + dhcp => $Hosts->{$host}{dhcp}, dumpExpire => 1}); + } else { + # + # Queue regular background backup + # + unshift(@BgQueue, + {host => $host, user => "BackupPC", reqTime => time, + dhcp => $Hosts->{$host}{dhcp}}); + } + $BgQueueOn{$host} = 1; + } + + return $skipped; +} + # # Queue all the hosts for backup. This means queuing all the fixed # ip hosts and all the dhcp address ranges. We also additionally @@ -1591,57 +1714,9 @@ sub HostSortCompare sub QueueAllPCs { my $nSkip = 0; + foreach my $host ( sort(HostSortCompare keys(%$Hosts)) ) { - delete($Status{$host}{backoffTime}) - if ( defined($Status{$host}{backoffTime}) - && $Status{$host}{backoffTime} < time ); - next if ( defined($Jobs{$host}) - || $BgQueueOn{$host} - || $UserQueueOn{$host} - || $CmdQueueOn{$host} ); - if ( $Hosts->{$host}{dhcp} ) { - $Status{$host}{dhcpCheckCnt}++; - if ( $RunNightlyWhenIdle ) { - # - # Once per night queue a check for DHCP hosts that just - # checks for expired dumps. We need to do this to handle - # the case when a DHCP host has not been on the network for - # a long time, and some of the old dumps need to be expired. - # Normally expiry checks are done by BackupPC_dump only - # after the DHCP hosts has been detected on the network. - # - unshift(@BgQueue, - {host => $host, user => "BackupPC", reqTime => time, - dhcp => 0, dumpExpire => 1}); - $BgQueueOn{$host} = 1; - } - } else { - # - # this is a fixed ip host: queue it - # - if ( $Info{DUlastValue} > $Conf{DfMaxUsagePct} ) { - # - # Since we are out of disk space, instead of queuing - # a regular job, queue an expire check instead. That - # way if the admin reduces the number of backups to - # keep then we will actually delete them. Otherwise - # BackupPC_dump will never run since we have exceeded - # the limit. - # - $nSkip++; - unshift(@BgQueue, - {host => $host, user => "BackupPC", reqTime => time, - dhcp => $Hosts->{$host}{dhcp}, dumpExpire => 1}); - } else { - # - # Queue regular background backup - # - unshift(@BgQueue, - {host => $host, user => "BackupPC", reqTime => time, - dhcp => $Hosts->{$host}{dhcp}}); - } - $BgQueueOn{$host} = 1; - } + $nSkip += QueueOnePC($host); } if ( $nSkip ) { print(LOG $bpc->timeStamp, @@ -1748,7 +1823,7 @@ sub catch_signal close(LOG); LogFileOpen(); print(LOG "Fatal error: unhandled signal $SigName\n"); - unlink("$TopDir/log/BackupPC.pid"); + unlink("$LogDir/BackupPC.pid"); confess("Got new signal $SigName... quitting\n"); } else { $SigName = shift; @@ -1760,9 +1835,9 @@ sub catch_signal # sub LogFileOpen { - mkpath("$TopDir/log", 0, 0777) if ( !-d "$TopDir/log" ); - open(LOG, ">>$TopDir/log/LOG") - || die("Can't create LOG file $TopDir/log/LOG"); + mkpath($LogDir, 0, 0777) if ( !-d $LogDir ); + open(LOG, ">>$LogDir/LOG") + || die("Can't create LOG file $LogDir/LOG"); close(STDOUT); close(STDERR); open(STDOUT, ">&LOG"); @@ -1787,7 +1862,7 @@ sub ServerSocketInit print(LOG $bpc->timeStamp, "unix socket() failed: $!\n"); exit(1); } - my $sockFile = "$TopDir/log/BackupPC.sock"; + my $sockFile = "$LogDir/BackupPC.sock"; unlink($sockFile); if ( !bind(SERVER_UNIX, sockaddr_un($sockFile)) ) { print(LOG $bpc->timeStamp, "unix bind() failed: $!\n"); @@ -1870,7 +1945,7 @@ sub ServerShutdown } delete($Info{pid}); StatusWrite(); - unlink("$TopDir/log/BackupPC.pid"); + unlink("$LogDir/BackupPC.pid"); exit(1); }