jobd/gc.sh

   1 #!/bin/sh
   2
   3 # NOTE: additional options can be passed to git repack by specifying
   4 #       them after the project name, for example:
   5 #         gc.sh my-project -f
   6
   7 . @basedir@/shlib.sh
   8
   9 set -e
  10 trap 'if [ $? != 0 ]; then echo "gc failed dir: $PWD" >&2; fi' EXIT
  11
  12 # packing options
  13 packopts='--window=50 --window-memory=1g --depth=50'
  14
  15 umask 002
  16 [ "$cfg_permission_control" != "Hooks" ] || umask 000
  17
  18 pidactive() {
  19         if _result="$(kill -0 "$1" 2>&1)"; then
  20                 # process exists and we have permission to signal it
  21                 return 0
  22         fi
  23         case "$_result" in *"not permitted"*)
  24                 # we do not have permission to signal the process
  25                 return 0
  26         esac
  27         # process does not exist
  28         return 1
  29 }
  30
  31 createlock() {
  32         # A .lock file should only exist for much less than a second.
  33         # If we see a stale lock file (> 1h old), remove it and then,
  34         # just in case, wait 30 seconds for any process whose .lock
  35         # we might have just removed (it's racy) to finish doing what
  36         # should take much less than a second to do.
  37         _stalelock="$(find "$1.lock" -maxdepth 1 -mmin +60 -print 2>/dev/null || :)"
  38         if [ -n "$_stalelock" ]; then
  39                 rm -f "$_stalelock"
  40                 sleep 30
  41         fi
  42         for _try in p p n; do
  43                 if (set -C; > "$1.lock") 2>/dev/null; then
  44                         echo "$1.lock"
  45                         return 0
  46                 fi
  47                 # delay and try again
  48                 [ "$_try" != "p" ] || sleep 1
  49         done
  50         # cannot create lock file
  51         return 1
  52 }
  53
  54 # return true if there's more than one objects/pack-<sha>.pack file or
  55 # ANY sha-1 files in objects
  56 is_dirty() {
  57         _packs=$(find objects/pack -type f -name "pack-$octet20.pack" -print | wc -l)
  58         if [ $_packs != 1 ] && [ $_packs != 0 ]; then
  59                 return 0
  60         fi
  61         _objs=$(find objects/$octet -type f -name "$octet19" -print 2>/dev/null | wc -l)
  62         [ $_objs -ne 0 ]
  63 }
  64
  65 # if the current directory is_gfi_mirror then repack all packs listed in gfi-packs
  66 repack_gfi_packs() {
  67         is_gfi_mirror || return 0
  68         [ -s gfi-packs ] || return 0
  69         while IFS=': ' read -r _pack _junk; do
  70                 if [ -s "$_pack" -a -s "${_pack%.pack}.idx" ]; then
  71                         git show-index < "${_pack%.pack}.idx" | cut -d ' ' -f 2
  72                 fi
  73         done < gfi-packs | \
  74         git pack-objects $packopts --no-reuse-delta --delta-base-offset \
  75                 --non-empty --threads=1 $quiet objects/pack/packtmp | \
  76         while read -r _newpack; do
  77                 rm -f objects/pack/pack-$_newpack.*
  78                 ln -f objects/pack/packtmp-$_newpack.pack objects/pack/pack-$_newpack.pack
  79                 ln -f objects/pack/packtmp-$_newpack.idx objects/pack/pack-$_newpack.idx
  80                 rm -f objects/pack/packtmp-$_newpack.*
  81         done
  82         rm -f gfi-packs
  83 }
  84
  85 # HEADSHA="$(pack_is_complete /full/path/to/some.pack /full/path/to/packed-refs "$(cat HEAD)")"
  86 pack_is_complete() {
  87         # Must have a matching .idx file and a non-empty packed-refs file
  88         [ -s "${1%.pack}.idx" ] || return 1
  89         [ -s "$2" ] || return 1
  90         _headsha=
  91         case "$3" in
  92                 $octet20)
  93                         _headsha="$3"
  94                         ;;
  95                 "ref: refs/"?*|"ref:refs/"?*|"refs/"?*)
  96                         _headmatch="${3#ref:}"
  97                         _headmatch="${_headmatch# }"
  98                         _headmatchpat="$(echo "$_headmatch" | sed -e 's/\([.$]\)/\\\1/g')"
  99                         _headsha="$(grep -e "^$octet20 $_headmatchpat\$" < "$2" | \
 100                                     cut -d ' ' -f 1)"
 101                         case "$_headsha" in $octet20) :;; *)
 102                                 return 1
 103                         esac
 104                         ;;
 105                 *)
 106                         # bad HEAD
 107                         return 1
 108         esac
 109         rm -rf pack_is_complete_test
 110         mkdir pack_is_complete_test
 111         mkdir pack_is_complete_test/refs
 112         mkdir pack_is_complete_test/objects
 113         mkdir pack_is_complete_test/objects/pack
 114         echo "$_headsha" > pack_is_complete_test/HEAD
 115         ln -s "$1" pack_is_complete_test/objects/pack/
 116         ln -s "${1%.pack}.idx" pack_is_complete_test/objects/pack/
 117         ln -s "$2" pack_is_complete_test/packed-refs
 118         _count="$(git --git-dir=pack_is_complete_test rev-list --count --all 2>/dev/null || :)"
 119         rm -rf pack_is_complete_test
 120         [ -n "$_count" ] || return 1
 121         [ "$_count" -gt 0 ] 2>/dev/null || return 1
 122         echo "$_headsha"
 123 }
 124
 125 proj="$1"
 126 shift
 127 cd "$cfg_reporoot/$proj.git"
 128
 129 # date -R is linux-only, POSIX equivalent is '+%a, %d %b %Y %T %z'
 130 datefmt='+%a, %d %b %Y %T %z'
 131
 132 if check_interval lastgc $cfg_min_gc_interval; then
 133         progress "= [$proj] garbage check skip (last at $(config_get lastgc))"
 134         exit 0
 135 fi
 136 if [ -e .nogc ]; then
 137         progress "x [$proj] garbage check disabled"
 138         exit 0
 139 fi
 140
 141 # Avoid unnecessary garbage collections:
 142 #   1. If lastreceive is set and is older than lastgc
 143 #   -AND-
 144 #   2. We are not a fork (! -s alternates) -OR- lastparentgc is older than lastgc
 145
 146 # If lastgc is NOT set or lastreceive is NOT set we MUST run gc
 147 # If we are a fork and lastparentgc is NOT set we MUST run gc
 148
 149 # If the repo is dirty after removing any crud we MUST run gc
 150
 151 gcstart="$(date "$datefmt")"
 152 skipgc=
 153 isfork=
 154 [ -s objects/info/alternates ] && isfork=1
 155 lastparentgcsecs=
 156 [ -n "$isfork" ] && lastparentgcsecs="$(config_get_date_seconds lastparentgc || :)"
 157 lastreceivesecs=
 158 if lastreceivesecs="$(config_get_date_seconds lastreceive)" && \
 159    lastgcsecs="$(config_get_date_seconds lastgc)" && \
 160    [ $lastreceivesecs -lt $lastgcsecs ]; then
 161         # We've run gc since we last received, so maybe we can skip,
 162         # check if not fork or fork and lastparentgc < lastgc
 163         if [ -n "$isfork" ]; then
 164                 if [ -n "$lastparentgcsecs" ] && \
 165                    [ $lastparentgcsecs -lt $lastgcsecs ]; then
 166                         # We've run gc since our parent ran gc so we can skip
 167                         skipgc=1
 168                 fi
 169         else
 170                 # We don't have any alternates (we're not a forK) so we can skip
 171                 skipgc=1
 172         fi
 173 fi
 174
 175 # be compatibile with gc.pid file from newer Git releases
 176
 177 hn="$(hostname)"
 178 lockf=gc.pid
 179 active=
 180 if [ "$(createlock "$lockf")" ]; then
 181         # If $lockf is:
 182         #   1) less than 12 hours old
 183         #   2) contains two fields (pid hostname) NO trailing NL
 184         #   3) the hostname is different OR the pid is still alive
 185         # then we exit as another active process is holding the lock
 186         if [ "$(find "$lockf" -maxdepth 1 -mmin -720 -print 2>/dev/null)" ]; then
 187                 apid=
 188                 ahost=
 189                 read -r apid ahost ajunk < "$lockf" || :
 190                 if [ "$apid" ] && [ "$ahost" ]; then
 191                         if [ "$ahost" != "$hn" ] || pidactive "$apid"; then
 192                                 active=1
 193                         fi
 194                 fi
 195         fi
 196 else
 197         echo >&2 "[$proj] unable to create gc.pid.lock file"
 198         exit 1
 199 fi
 200 if [ -n "$active" ]; then
 201         rm -f "$lockf.lock"
 202         echo >&2 "[$proj] gc already running on machine '$ahost' pid '$apid'"
 203         exit 1
 204 fi
 205 printf "%s %s" "$$" "$hn" > "$lockf.lock"
 206 chmod 0664 "$lockf.lock"
 207 mv -f "$lockf.lock" "$lockf"
 208
 209 # Remove any stale pack remnants that are more than an hour old.
 210 # Stale pack fragments are defined as any pack-<sha1>.ext where .ext is NOT
 211 # .pack AND the corresponding .pack DOES NOT exist.  A bunch of stale
 212 # pack-<sha1>.idx files without their corresponding .pack files are worthless
 213 # and just waste space.  Normally there shouldn't be any remnants but actually
 214 # this can happen when things are interrupted at just the wrong time.
 215 # Note that the objects/pack directory is created by git init and should
 216 # always exist.
 217 find objects/pack -maxdepth 1 -type f -mmin +60 -name "pack-$octet20.?*" -print | \
 218 sed -e 's/^objects\/pack\/pack-//; s/\..*$//' | LC_ALL=C sort -u | \
 219 while read packsha; do
 220         [ ! -e "objects/pack/pack-$packsha.pack" ] || continue
 221         rm -f "objects/pack/pack-$packsha".?*
 222 done
 223
 224 # Remove any stale pack .keep files that are more than 12 hours old.
 225 # We don't do anything to create any permanent pack .keep files, so they must
 226 # be remnants from some failed push or something.  Removing the .keep will
 227 # allow the pack to be properly repacked.
 228 find objects/pack -maxdepth 1 -type f -mmin +720 -name "pack-$octet20.keep" -print | \
 229 while read packkeep; do
 230         rm -f "$packkeep"
 231 done
 232
 233 # Remove any stale tmp_pack_* or tmp_idx_* or packtmp-* files that are more than 12 hours old.
 234 find objects/pack -maxdepth 1 -type f -mmin +720 -name "tmp_pack_?*" -print | \
 235 while read packtmp; do
 236         rm -f "$packtmp"
 237 done
 238 find objects/pack -maxdepth 1 -type f -mmin +720 -name "tmp_idx_?*" -print | \
 239 while read packtmp; do
 240         rm -f "$packtmp"
 241 done
 242 find objects/pack -maxdepth 1 -type f -mmin +720 -name "packtmp-?*" -print | \
 243 while read packtmp; do
 244         rm -f "$packtmp"
 245 done
 246
 247 # Remove any stale *.temp files in the objects area that are more than 12 hours old.
 248 # This can be stale sha1.temp, or stale *.pack.temp so we kill all stale *.temp.
 249 find objects -type f -mmin +720 -name "*.temp" -print | \
 250 while read objtmp; do
 251         rm -f "$objtmp"
 252 done
 253
 254 # Remove any stale git-svn temp files that are more than 12 hours old.
 255 # The git-svn process creates temp files with random 10 character names
 256 # in the root of $GIT_DIR.  Unfortunately they do not have a recognizable
 257 # prefix, so we just have to kill any files with a 10-character name.  We
 258 # do this only for git-svn mirrors.  All characters are chosen from
 259 # [A-Za-z0-9_] so we can at least check that and fortunately the only
 260 # collision is 'FETCH_HEAD' but that shouldn't matter.
 261 # There may also be temp files with a Git_ prefix as well.
 262 if is_svn_mirror; then
 263         _randchar='[A-Za-z0-9_]'
 264         _randchar2="$_randchar$_randchar"
 265         _randchar4="$_randchar2$_randchar2"
 266         _randchar10="$_randchar4$_randchar4$_randchar2"
 267         find . -maxdepth 1 -type f -mmin +720 -name "$_randchar10" -print | \
 268         while read tmpcrud; do
 269                 rm -f "$tmpcrud"
 270         done
 271         find . -maxdepth 1 -type f -mmin +720 -name "Git_*" -print | \
 272         while read tmpcrud; do
 273                 rm -f "$tmpcrud"
 274         done
 275 fi
 276
 277 # Remove any stale fast_import_crash_<pid> files that are more than 3 days old.
 278 if is_gfi_mirror; then
 279         find . -maxdepth 1 -type f -mmin +4320 -name "fast_import_crash_?*" -print | \
 280         while read fastcrash; do
 281                 rm -f "$fastcrash"
 282         done
 283 fi
 284
 285 # Do not skip gc if the repo is dirty
 286 if [ -n "$skipgc" ] && ! is_dirty; then
 287         progress "= [$proj] garbage check nothing but crud removal to do (`date`)"
 288         config_set lastgc "$gcstart"
 289         rm -f "$lockf"
 290         exit 0
 291 fi
 292
 293 bumptime=
 294 if [ -n "$isfork" ] && [ -z "$lastparentgcsecs" ]; then
 295         # set lastparentgc and then update gcstart to be at least 1 second later
 296         config_set lastparentgc "$gcstart"
 297         bumptime=1
 298 fi
 299 if [ -z "$lastreceivesecs" ]; then
 300         # set lastreceive and then update gcstart to be at least 1 second later
 301         config_set lastreceive "$gcstart"
 302         bumptime=1
 303 fi
 304 if [ -n "$bumptime" ]; then
 305         sleep 1
 306         gcstart="$(date "$datefmt")"
 307 fi
 308
 309 progress "+ [$proj] garbage check (`date`)"
 310
 311 # safe pruning: we put all our objects in all forks, then we can
 312 # safely get rid of extra ones; repacks in forks will get rid of
 313 # the redundant ones again then
 314 forkdir="$proj"
 315 if [ -d "../${forkdir##*/}" ]; then
 316         # It is enough to copy objects just one level down and get_repo_list
 317         # takes a regular expression (which is automatically prefixed with '^')
 318         # so we can easily match forks exactly one level down from this project
 319         get_repo_list "$forkdir/[^/]*:" |
 320                 while read fork; do
 321                         # Ignore forks that do not exist or are symbolic links
 322                         [ ! -L "$cfg_reporoot/$fork.git" -a -d "$cfg_reporoot/$fork.git" ] || \
 323                                 continue
 324                         # Or do not have a non-zero length alternates file
 325                         [ -s "$cfg_reporoot/$fork.git/objects/info/alternates" ] || \
 326                                 continue
 327                         # Match objects in parent project
 328                         for d in objects/?? objects/pack; do
 329                                 [ "$d" != "objects/??" ] || continue
 330                                 mkdir -p "$cfg_reporoot/$fork.git/$d"
 331                                 [ "$d" != "objects/pack" ] || \
 332                                         [ "$(echo objects/pack/*)" != \
 333                                                         "objects/pack/*" ] || \
 334                                         continue
 335                                 ln -f "$d"/* "$cfg_reporoot/$fork.git/$d" || :
 336                         done
 337                         # Update the fork's lastparentgc date (must be current, not $gcstart)
 338                         GIT_DIR="$cfg_reporoot/$fork.git" git config \
 339                                 gitweb.lastparentgc "$(date "$datefmt")"
 340                 done
 341 fi
 342
 343 quiet=; [ -n "$show_progress" ] || quiet=-q
 344
 345 git pack-refs --all
 346 repack_gfi_packs
 347 rm -f bundles/*
 348 rm -f objects/pack/pack-*.bndl
 349 git repack $packopts -a -d -l $quiet "$@"
 350 allpacks="$(echo objects/pack/pack-$octet20.pack)"
 351 curhead="$(cat HEAD)"
 352 pkrf=
 353 [ ! -e packed-refs ] || pkrf=packed-refs
 354 eval "reposizek=$(( $(echo 0 $(du -k $pkrf $allpacks 2>/dev/null | awk '{print $1}') | \
 355         sed -e 's/ / + /g') ))"
 356 git prune
 357 git update-server-info
 358
 359 # darcs:// mirrors have a xxx.log file that will grow endlessly
 360 # if this is a mirror and the file exists, shorten it to 10000 lines
 361 # also take this opportunity to optimize the darcs repo
 362 if [ ! -e .nofetch ] && [ -n "$cfg_mirror" ]; then
 363         url="$(config_get baseurl || :)"
 364         case "$url" in darcs://*)
 365                 if [ -n "$cfg_mirror_darcs" ]; then
 366                         url="${url%/}"
 367                         basedarcs="$(basename "${url#darcs:/}")"
 368                         if [ -f "$basedarcs.log" ]; then
 369                                 tail -n 10000 "$basedarcs.log" > "$basedarcs.log.$$"
 370                                 mv -f "$basedarcs.log.$$" "$basedarcs.log"
 371                         fi
 372                         if [ -d "$basedarcs.darcs" ]; then
 373                                 (
 374                                         cd "$basedarcs.darcs"
 375                                         # Note that this does not optimize _darcs/inventories/ :(
 376                                         darcs optimize
 377                                 )
 378                         fi
 379                 fi
 380         esac
 381 fi
 382
 383 # Create a matching .bndl header file for the all-in-one pack we just created
 384 # but only if we're not a fork (otherwise the bundle would not be complete)
 385 if [ ! -s objects/info/alternates ]; then
 386         # There should only be one pack in $allpacks but if there was a
 387         # simultaneous push...
 388         # The one we just created will have a .idx and will NOT have a .keep
 389         pkfound=
 390         pkhead=
 391         for pk in $allpacks; do
 392                 [ -s "$pk" ] || continue
 393                 pkbase="${pk%.pack}"
 394                 [ -s "$pkbase.idx" ] || continue
 395                 [ ! -e "$pkbase.keep" ] || continue
 396                 if pkhead="$(pack_is_complete "$PWD/$pk" "$PWD/packed-refs" "$curhead")"; then
 397                         pkfound="$pkbase"
 398                         break;
 399                 fi
 400         done
 401         if [ -n "$pkfound" -a -n "$pkhead" ]; then
 402                 {
 403                         echo "# v2 git bundle"
 404                         sed -ne "/^$octet20 refs\/[^    ]*\$/ p" < packed-refs
 405                         echo "$pkhead HEAD"
 406                         echo ""
 407                 } > "$pkbase.bndl"
 408                 bndlsha="${pkbase#objects/pack/pack-}"
 409                 bndlshatrailer="${bndlsha#????????}"
 410                 bndlshaprefix="${bndlsha%$bndlshatrailer}"
 411                 bndlname="$(TZ=UTC date +%Y%m%d_%H%M%S)-${bndlshaprefix:-0}"
 412                 [ -d bundles ] || mkdir bundles
 413                 echo "$bndlsha" > "bundles/$bndlname"
 414         fi
 415 fi
 416
 417 # Record the size of this repo as the sum of its *.pack sizes as 1024-byte blocks
 418 config_set_raw girocco.reposizek "${reposizek:-0}"
 419
 420 # We use $gcstart here to avoid a race where a push occurs during the gc itself
 421 # and the next future gc could be incorrectly skipped if we used the current
 422 # timestamp here instead
 423 config_set lastgc "$gcstart"
 424 rm -f "$lockf"
 425
 426 progress "- [$proj] garbage check (`date`)"