git-cache-proxy: introduce lockfile

[chiark-utils.git] / scripts / git-cache-proxy

diff --git a/scripts/git-cache-proxy b/scripts/git-cache-proxy
index f797e74920cc36822141ad2422ad683d6f8e4281..b4b23001cb3ed42686ccaf0c1392c7f4f0bc40c5 100755 (executable)
--- a/scripts/git-cache-proxy
+++ b/scripts/git-cache-proxy
@@ -40,6 +40,7 @@ use POSIX;
 use Socket;
 use Sys::Syslog;
 use Fcntl qw(:flock SEEK_SET);
+use File::Path qw(remove_tree);
 
 our $us = 'git-cache-proxy';
 
@@ -126,6 +127,8 @@ sub gitfail ($) {
 
 #---------- argument parsing ----------
 
+our $housekeepingthreshdays = 1;
+our $treeexpiredays = 21;
 our $fetchtimeout = 1800;
 our $maxfetchtimeout = 3600;
 our $cachedir = '/var/cache/git-cache-proxy';
@@ -152,13 +155,32 @@ for (;;) {
 
 !@ARGV or fail "bad usage: no non-option arguments permitted";
 
-#---------- main program ----------
-
-chdir $cachedir or fail "chdir $cachedir: $!";
+#---------- utility functions ----------
 
-our ($service,$specpath,$spechost,$subdir);
-our ($tmpd,$gitd,$lock);
-our ($fetch,$url);
+sub lockfile ($$) {
+    my ($fh, $fn, $flockmode) = @_;
+    for (;;) {
+       close $fh;
+       open $fh, '+>', $fn or fail "open/create $fn for lock: $!";
+       if (!flock $fh, $lockmode) {
+           if ($lockmode & LOCK_NB and $! == EWOULDBLOCK) {
+               return 0; # ok then
+           }
+           fail "lock $fn".
+               (($flockmode & ~LOCK_NB) == LOCK_SH ? " (shared)" : "").
+               ": $!";
+       }
+       stat $fh or fail "stat opened $fn: $!";
+       my $fh_ino = ((stat _)[1]);
+       if (!stat $fn) {
+           $! == ENOENT or fail "stat $fn: $!";
+           next;
+       }
+       my $fn_ino = ((stat _)[1]);
+       return 1 if $fn_ino == $fh_ino;
+       # oh dear
+    }
+}
 
 sub xread {
     my $length = shift;
@@ -172,6 +194,19 @@ sub xread {
     return $buffer;
 }
 
+#---------- main program ----------
+
+chdir $cachedir or fail "chdir $cachedir: $!";
+
+our ($service,$specpath,$spechost,$subdir);
+our ($tmpd,$gitd,$lock);
+our ($fetch,$url);
+
+sub servinfo ($) {
+    my ($msg) = @_;
+    logm 'info', "service `$specpath': $msg";
+}
+
 sub readcommand () {
     $SIG{ALRM} = sub { fail "timeout" };
     alarm 30;
@@ -218,19 +253,18 @@ sub readcommand () {
     $subdir =~ s|,|\\,|g;
     $subdir =~ s|/|,|g;
 
-    logm 'info', "$specpath locking";
-    
     $tmpd= "$subdir\\.tmp";
     $gitd= "$subdir\\.git";
     $lock = "$subdir\\.lock";
+
+    servinfo "locking";
 }
 
 sub clonefetch () {
-    open LOCK, "+>", $lock or fail "open/create $lock: $!";
-    flock LOCK, LOCK_EX or fail "lock exclusive $lock: $!";
+    lockfile \*LOCK, $lock, LOCK_EX;
 
-    my $exists = stat $gitd;
-    $exists or $!==ENOENT or fail "stat $gitd: $!";
+    my $exists = lstat $gitd;
+    $exists or $!==ENOENT or fail "lstat $gitd: $!";
 
     our $fetchfail = '';
 
@@ -241,10 +275,10 @@ sub clonefetch () {
        if (!$exists) {
            system qw(rm -rf --), $tmpd;
            @cmd = (qw(git clone -q --mirror), $url, $tmpd);
-           logm 'info', "$specpath cloning @cmd";
+           servinfo "cloning";
        } else {
            @cmd = (qw(git remote update --prune));
-           logm 'info', "$specpath fetching @cmd";
+           servinfo "fetching";
        }
        my $cmd = "@cmd[0..1]";
 
@@ -263,10 +297,9 @@ sub clonefetch () {
        my $timedout = 0;
        {
            local $SIG{ALRM} = sub {
-               logm 'info', "$specpath fetch/clone timeout";
+               servinfo "fetch/clone timeout";
                $timedout=1; kill 9, -$child;
            };
-logm 'info', "timeout=$fetchtimeout";
            alarm($fetchtimeout);
            $!=0; { local $/=undef; $fetcherr = <FETCHERR>; }
            !FETCHERR->error or fail "read pipe from fetch/clone: $!";
@@ -289,7 +322,7 @@ logm 'info', "timeout=$fetchtimeout";
            if ($fetch >= 2) {
                gitfail $fetchfail;
            } else {
-               logm 'info', "$specpath fetch/clone failed: $fetchfail";
+               servinfo "fetch/clone failed: $fetchfail";
            }
        }
 
@@ -305,11 +338,8 @@ logm 'info', "timeout=$fetchtimeout";
        gitfail "no cached data, and not cloned: $fetchfail";
     }
 
-    logm 'info', "$specpath sharing";
-    flock LOCK, LOCK_UN or fail "unlock $lock: $!";
-    flock LOCK, LOCK_SH or fail "lock shared $lock: $!";
-    # actually, just relocking as shared would have the same semantics
-    # but it's best to be explicit
+    servinfo "sharing";
+    lockfile \*LOCK, $lock, LOCK_SH; # NB releases and relocks
 
     if (chdir $gitd) {
        return 1;
@@ -321,8 +351,66 @@ logm 'info', "timeout=$fetchtimeout";
     return 0;
 }
 
+sub housekeeping () {
+    foreach $lock (<[a-z]*\\.lock>) {
+       if (!lstat $lock) {
+           $! == ENOENT or fail "housekeeping: $lock: lstat: $!";
+           next;
+       }
+       if (-M _ <= $treeexpiredays) {
+           logm 'debug', "housekeeping: $lock: not too old";
+           next;
+       }
+       my $subdir = $lock;  $subdir =~ s/\\.lock$//;
+       my $ok = 1;
+       foreach my $suffix (qw(tmp git)) {
+           my $dir = "${subdir}\\.$suffix";
+           my $errs;
+           remove_tree($dir, { safe=>1, error=>\$errs });
+           $ok = 0 if @$errs;
+           foreach my $err (@$errs) {
+               logm 'warning', "problem deleting: $err[0]: $err[1]";
+           }
+       }
+       if ($ok) {
+           
+
+sub housekeepingcheck ($$) {
+    my ($dofork, $force) = @_;
+       or fail "open/create Housekeeping.lock: $!";
+    if (!$force) {
+       if (!lockfile \*HLOCK, "Housekeeping.lock", LOCK_EX|LOCK_NB) {
+           logm 'debug', "housekeeping lock taken, not running";
+           close HLOCK;
+           return 0;
+       }
+    }
+    if ($force) {
+       logm 'info', "housekeeping forced";
+    } elsif (!lstat "Housekeeping.stamp") {
+       $! == ENOENT or fail "lstat Housekeeping.stamp: $!";
+       logm 'info', "housekeeping stamp missing, will run";
+    } elsif (-M _ <= $housekeepingthreshdays) {
+       logm 'debug', "housekeeping done recently";
+       close HLOCK;
+       return 0;
+    }
+    if ($dofork) {
+       my $child = fork;
+       defined $child or fail "fork for housekeeping: $!";
+       if (!$child) {
+           housekeeping();
+           exit 0;
+       }
+       return 1;
+    } else {
+       housekeeping();
+       return 1;
+    }
+}
+
 sub runcommand () {
-    logm 'info', "$specpath servicing";
+    servinfo "servicing";
     exec qw(git-upload-pack --strict --timeout=1000 .)
        or fail "exec git-upload-pack: $!";
 }