jobd/gc.sh

   1 #!/bin/sh
   2
   3 . @basedir@/shlib.sh
   4
   5 set -e
   6 trap 'if [ $? != 0 ]; then echo "gc failed dir: $PWD" >&2; fi; rm -f "$bang_log"' EXIT
   7
   8 # packing options
   9 packopts='--window=50 --window-memory=1g --depth=50'
  10
  11 umask 002
  12 [ "$cfg_permission_control" != "Hooks" ] || umask 000
  13
  14 pidactive() {
  15         if _result="$(kill -0 "$1" 2>&1)"; then
  16                 # process exists and we have permission to signal it
  17                 return 0
  18         fi
  19         case "$_result" in *"not permitted"*)
  20                 # we do not have permission to signal the process
  21                 return 0
  22         esac
  23         # process does not exist
  24         return 1
  25 }
  26
  27 createlock() {
  28         for _try in p n; do
  29                 if (set -C; > "$1.lock") 2>/dev/null; then
  30                         echo "$1.lock"
  31                         return 0
  32                 fi
  33                 # delay and try again
  34                 [ "$_try" != "p" ] || sleep 1
  35         done
  36         # cannot create temp file
  37         return 1
  38 }
  39
  40 # if the current directory is_gfi_mirror then repack all packs listed in gfi-packs
  41 repack_gfi_packs() {
  42         is_gfi_mirror || return 0
  43         [ -s gfi-packs ] || return 0
  44         while IFS=': ' read -r _pack _junk; do
  45                 if [ -s "$_pack" -a -s "${_pack%.pack}.idx" ]; then
  46                         git show-index < "${_pack%.pack}.idx" | cut -d ' ' -f 2
  47                 fi
  48         done < gfi-packs | \
  49         git pack-objects $packopts --no-reuse-delta --delta-base-offset \
  50                 --non-empty --threads=1 $quiet objects/pack/packtmp | \
  51         while read -r _newpack; do
  52                 rm -f objects/pack/pack-$_newpack.*
  53                 ln -f objects/pack/packtmp-$_newpack.pack objects/pack/pack-$_newpack.pack
  54                 ln -f objects/pack/packtmp-$_newpack.idx objects/pack/pack-$_newpack.idx
  55                 rm -f objects/pack/packtmp-$_newpack.*
  56         done
  57         rm -f gfi-packs
  58 }
  59
  60 # HEADSHA="$(pack_is_complete /full/path/to/some.pack /full/path/to/packed-refs "$(cat HEAD)")"
  61 pack_is_complete() {
  62         # Must have a matching .idx file and a non-empty packed-refs file
  63         [ -s "${1%.pack}.idx" ] || return 1
  64         [ -s "$2" ] || return 1
  65         _headsha=
  66         case "$3" in
  67                 $octet20)
  68                         _headsha="$3"
  69                         ;;
  70                 "ref: refs/"?*|"ref:refs/"?*|"refs/"?*)
  71                         _headmatch="${3#ref:}"
  72                         _headmatch="${_headmatch# }"
  73                         _headmatchpat="$(echo "$_headmatch" | sed -e 's/\([.$]\)/\\\1/g')"
  74                         _headsha="$(grep -e "^$octet20 $_headmatchpat\$" < "$2" | \
  75                                     cut -d ' ' -f 1)"
  76                         case "$_headsha" in $octet20) :;; *)
  77                                 return 1
  78                         esac
  79                         ;;
  80                 *)
  81                         # bad HEAD
  82                         return 1
  83         esac
  84         rm -rf pack_is_complete_test
  85         mkdir pack_is_complete_test
  86         mkdir pack_is_complete_test/refs
  87         mkdir pack_is_complete_test/objects
  88         mkdir pack_is_complete_test/objects/pack
  89         echo "$_headsha" > pack_is_complete_test/HEAD
  90         ln -s "$1" pack_is_complete_test/objects/pack/
  91         ln -s "${1%.pack}.idx" pack_is_complete_test/objects/pack/
  92         ln -s "$2" pack_is_complete_test/packed-refs
  93         _count="$(git --git-dir=pack_is_complete_test rev-list --count --all 2>/dev/null || :)"
  94         rm -rf pack_is_complete_test
  95         [ -n "$_count" ] || return 1
  96         [ "$_count" -gt 0 ] 2>/dev/null || return 1
  97         echo "$_headsha"
  98 }
  99
 100 proj="$1"
 101 cd "$cfg_reporoot/$proj.git"
 102
 103 # date -R is linux-only, POSIX equivalent is '+%a, %d %b %Y %T %z'
 104 datefmt='+%a, %d %b %Y %T %z'
 105
 106 if check_interval lastgc $cfg_min_gc_interval; then
 107         progress "= [$proj] garbage check skip (last at $(config_get lastgc))"
 108         exit 0
 109 fi
 110 if [ -e .nogc ]; then
 111         progress "x [$proj] garbage check disabled"
 112         exit 0
 113 fi
 114
 115 # Avoid unnecessary garbage collections:
 116 #   1. If lastreceive is set and is older than lastgc
 117 #   -AND-
 118 #   2. We are not a fork (! -s alternates) -OR- lastparentgc is older than lastgc
 119
 120 # If lastgc is NOT set or lastreceive is NOT set we MUST run gc
 121 # If we are a fork and lastparentgc is NOT set we MUST run gc
 122
 123 gcstart="$(date "$datefmt")"
 124 skipgc=
 125 isfork=
 126 [ -s objects/info/alternates ] && isfork=1
 127 lastparentgcsecs=
 128 [ -n "$isfork" ] && lastparentgcsecs="$(config_get_date_seconds lastparentgc || :)"
 129 lastreceivesecs=
 130 if lastreceivesecs="$(config_get_date_seconds lastreceive)" && \
 131    lastgcsecs="$(config_get_date_seconds lastgc)" && \
 132    [ $lastreceivesecs -lt $lastgcsecs ]; then
 133         # We've run gc since we last received, so maybe we can skip,
 134         # check if not fork or fork and lastparentgc < lastgc
 135         if [ -n "$isfork" ]; then
 136                 if [ -n "$lastparentgcsecs" ] && \
 137                    [ $lastparentgcsecs -lt $lastgcsecs ]; then
 138                         # We've run gc since our parent ran gc so we can skip
 139                         skipgc=1
 140                 fi
 141         else
 142                 # We don't have any alternates (we're not a forK) so we can skip
 143                 skipgc=1
 144         fi
 145 fi
 146
 147 # be compatibile with gc.pid file from newer Git releases
 148
 149 hn="$(hostname)"
 150 lockf=gc.pid
 151 active=
 152 if [ "$(createlock "$lockf")" ]; then
 153         # If $lockf is:
 154         #   1) less than 12 hours old
 155         #   2) contains two fields (pid hostname) NO trailing NL
 156         #   3) the hostname is different OR the pid is still alive
 157         # then we exit as another active process is holding the lock
 158         if [ "$(find "$lockf" -mmin -720 -print 2>/dev/null)" ]; then
 159                 apid=
 160                 ahost=
 161                 read -r apid ahost ajunk < "$lockf" || :
 162                 if [ "$apid" ] && [ "$ahost" ]; then
 163                         if [ "$ahost" != "$hn" ] || pidactive "$apid"; then
 164                                 active=1
 165                         fi
 166                 fi
 167         fi
 168 else
 169         echo >&2 "[$proj] unable to create gc.pid.lock file"
 170         exit 1
 171 fi
 172 if [ -n "$active" ]; then
 173         rm "$lockf.lock"
 174         echo >&2 "[$proj] gc already running on machine '$ahost' pid '$apid'"
 175         exit 1
 176 fi
 177 printf "%s %s" "$$" "$hn" > "$lockf.lock"
 178 chmod 0664 "$lockf.lock"
 179 mv -f "$lockf.lock" "$lockf"
 180
 181 if [ -n "$skipgc" ]; then
 182         progress "= [$proj] garbage check nothing to do (`date`)"
 183         config_set lastgc "$gcstart"
 184         rm "$lockf"
 185         exit 0
 186 fi
 187
 188 bumptime=
 189 if [ -n "$isfork" ] && [ -z "$lastparentgcsecs" ]; then
 190         # set lastparentgc and then update gcstart to be at least 1 second later
 191         config_set lastparentgc "$gcstart"
 192         bumptime=1
 193 fi
 194 if [ -z "$lastreceivesecs" ]; then
 195         # set lastreceive and then update gcstart to be at least 1 second later
 196         config_set lastreceive "$gcstart"
 197         bumptime=1
 198 fi
 199 if [ -n "$bumptime" ]; then
 200         sleep 1
 201         gcstart="$(date "$datefmt")"
 202 fi
 203
 204 progress "+ [$proj] garbage check (`date`)"
 205
 206 # Remove any stale pack remnants that are more than an hour old.
 207 # Stale pack fragments are defined as any pack-<sha1>.ext where .ext is NOT
 208 # .pack AND the corresponding .pack DOES NOT exist.  A bunch of stale
 209 # pack-<sha1>.idx files without their corresponding .pack files are worthless
 210 # and just waste space.  Normally there shouldn't be any remnants but actually
 211 # this can happen when things are interrupted at just the wrong time.
 212 # Note that the objects/pack directory is created by git init and should
 213 # always exist.
 214 find objects/pack -maxdepth 1 -type f -mmin +60 -name "pack-$octet20.?*" | \
 215 sed -e 's/^objects\/pack\/pack-//; s/\..*$//' | sort -u | \
 216 while read packsha; do
 217         [ ! -e "objects/pack/pack-$packsha.pack" ] || continue
 218         rm -f "objects/pack/pack-$packsha".?*
 219 done
 220
 221 # safe pruning: we put all our objects in all forks, then we can
 222 # safely get rid of extra ones; repacks in forks will get rid of
 223 # the redundant ones again then
 224 forkdir="$1"
 225 if [ -d "../${forkdir##*/}" ]; then
 226         # It is enough to copy objects just one level down and get_repo_list
 227         # takes a regular expression (which is automatically prefixed with '^')
 228         # so we can easily match forks exactly one level down from this project
 229         get_repo_list "$forkdir/[^/]*:" |
 230                 while read fork; do
 231                         # Ignore forks that do not exist or are symbolic links
 232                         [ ! -L "$cfg_reporoot/$fork.git" -a -d "$cfg_reporoot/$fork.git" ] || \
 233                                 continue
 234                         # Or do not have a non-zero length alternates file
 235                         [ -s "$cfg_reporoot/$fork.git/objects/info/alternates" ] || \
 236                                 continue
 237                         # Match objects in parent project
 238                         for d in objects/?? objects/pack; do
 239                                 [ "$d" != "objects/??" ] || continue
 240                                 mkdir -p "$cfg_reporoot/$fork.git/$d"
 241                                 [ "$d" != "objects/pack" ] || \
 242                                         [ "$(echo objects/pack/*)" != \
 243                                                         "objects/pack/*" ] || \
 244                                         continue
 245                                 ln -f "$d"/* "$cfg_reporoot/$fork.git/$d" || :
 246                         done
 247                         # Update the fork's lastparentgc date (must be current, not $gcstart)
 248                         GIT_DIR="$cfg_reporoot/$fork.git" git config \
 249                                 gitweb.lastparentgc "$(date "$datefmt")"
 250                 done
 251 fi
 252
 253 quiet=; [ -n "$show_progress" ] || quiet=-q
 254
 255 git pack-refs --all
 256 repack_gfi_packs
 257 rm -f bundles/*
 258 rm -f objects/pack/pack-*.bndl
 259 git repack $packopts -a -d -l $quiet
 260 allpacks="$(echo objects/pack/pack-$octet20.pack)"
 261 curhead="$(cat HEAD)"
 262 eval "reposizek=$(( $(echo 0 $(du -k $allpacks 2>/dev/null | awk '{print $1}') | \
 263         sed -e 's/ / + /g') ))"
 264 git prune
 265 git update-server-info
 266
 267 # darcs:// mirrors have a xxx.log file that will grow endlessly
 268 # if this is a mirror and the file exists, shorten it to 10000 lines
 269 # also take this opportunity to optimize the darcs repo
 270 if [ ! -e .nofetch ] && [ -n "$cfg_mirror" ]; then
 271         url="$(config_get baseurl || :)"
 272         case "$url" in darcs://*)
 273                 if [ -n "$cfg_mirror_darcs" ]; then
 274                         url="${url%/}"
 275                         basedarcs="$(basename "${url#darcs:/}")"
 276                         if [ -f "$basedarcs.log" ]; then
 277                                 tail -n 10000 "$basedarcs.log" > "$basedarcs.log.$$"
 278                                 mv -f "$basedarcs.log.$$" "$basedarcs.log"
 279                         fi
 280                         if [ -d "$basedarcs.darcs" ]; then
 281                                 (
 282                                         cd "$basedarcs.darcs"
 283                                         # Note that this does not optimize _darcs/inventories/ :(
 284                                         darcs optimize
 285                                 )
 286                         fi
 287                 fi
 288         esac
 289 fi
 290
 291 # Create a matching .bndl header file for the all-in-one pack we just created
 292 # but only if we're not a fork (otherwise the bundle would not be complete)
 293 if [ ! -s objects/info/alternates ]; then
 294         # There should only be one pack in $allpacks but if there was a
 295         # simultaneous push...
 296         # The one we just created will have a .idx and will NOT have a .keep
 297         pkfound=
 298         pkhead=
 299         for pk in $allpacks; do
 300                 [ -s "$pk" ] || continue
 301                 pkbase="${pk%.pack}"
 302                 [ -s "$pkbase.idx" ] || continue
 303                 [ ! -e "$pkbase.keep" ] || continue
 304                 if pkhead="$(pack_is_complete "$PWD/$pk" "$PWD/packed-refs" "$curhead")"; then
 305                         pkfound="$pkbase"
 306                         break;
 307                 fi
 308         done
 309         if [ -n "$pkfound" -a -n "$pkhead" ]; then
 310                 {
 311                         echo "# v2 git bundle"
 312                         sed -ne "/^$octet20 refs\/[^    ]*\$/ p" < packed-refs
 313                         echo "$pkhead HEAD"
 314                         echo ""
 315                 } > "$pkbase.bndl"
 316                 bndlsha="${pkbase#objects/pack/pack-}"
 317                 bndlshatrailer="${bndlsha#????????}"
 318                 bndlshaprefix="${bndlsha%$bndlshatrailer}"
 319                 bndlname="$(TZ=UTC date +%Y%m%d_%H%M%S)-${bndlshaprefix:-0}"
 320                 [ -d bundles ] || mkdir bundles
 321                 echo "$bndlsha" > "bundles/$bndlname"
 322         fi
 323 fi
 324
 325 # Record the size of this repo as the sum of its *.pack sizes as 1024-byte blocks
 326 config_set_raw girocco.reposizek "${reposizek:-0}"
 327
 328 # We use $gcstart here to avoid a race where a push occurs during the gc itself
 329 # and the next future gc could be incorrectly skipped if we used the current
 330 # timestamp here instead
 331 config_set lastgc "$gcstart"
 332 rm "$lockf"
 333
 334 progress "- [$proj] garbage check (`date`)"