jobd/update.sh

   1 #!/bin/sh
   2
   3 . @basedir@/shlib.sh
   4
   5 set -e
   6
   7 if [ $# -ne 1 ]; then
   8         echo "Usage: update.sh projname" >&2
   9         exit 1
  10 fi
  11
  12 # date -R is linux-only, POSIX equivalent is '+%a, %d %b %Y %T %z'
  13 datefmt='+%a, %d %b %Y %T %z'
  14
  15 # freshen_loose_objects full-sha ...
  16 # if "$n" is a loose object, set its modification time to now
  17 # otherwise silently do nothing with no error.  To facilitate conversion
  18 # of mirror projects to push projects we also add group write  permission.
  19 freshen_loose_objects() {
  20         _list=
  21         for _sha; do
  22                 _fn="${_sha#??}"
  23                 _shard="${_sha%$_fn}"
  24                 _list="$_list objects/$_shard/$_fn"
  25         done
  26         if [ -n "$_list" ]; then
  27                 chmod ug+w $_list 2>/dev/null || :
  28                 touch -c $_list 2>/dev/null || :
  29         fi
  30 }
  31
  32 # darcs fast-export | git fast-import with error handling
  33 git_darcs_fetch() (
  34         set_utf8_locale
  35         _err1=
  36         _err2=
  37         exec 3>&1
  38         { read -r _err1 || :; read -r _err2 || :; } <<-EOT
  39         $(
  40                 exec 4>&3 3>&1 1>&4 4>&-
  41                 {
  42                         _e1=0
  43                         "$cfg_basedir"/bin/darcs-fast-export \
  44                                 --export-marks="$(pwd)/dfe-marks" \
  45                                 --import-marks="$(pwd)/dfe-marks" "$1" 3>&- || _e1=$?
  46                         echo $_e1 >&3
  47                 } | \
  48                 {
  49                         _e2=0
  50                         git fast-import \
  51                                 --export-marks="$(pwd)/gfi-marks" \
  52                                 --export-pack-edges="$(pwd)/gfi-packs" \
  53                                 --import-marks="$(pwd)/gfi-marks" \
  54                                 --force 3>&- || _e2=$?
  55                         echo $_e2 >&3
  56                 }
  57         )
  58         EOT
  59         exec 3>&-
  60         [ "$_err1" = 0 -a "$_err2" = 0 ]
  61         return $?
  62 )
  63
  64 # bzr fast-export | git fast-import with error handling
  65 git_bzr_fetch() (
  66         set_utf8_locale
  67         BZR_LOG=/dev/null
  68         export BZR_LOG
  69         _err1=
  70         _err2=
  71         exec 3>&1
  72         { read -r _err1 || :; read -r _err2 || :; } <<-EOT
  73         $(
  74                 exec 4>&3 3>&1 1>&4 4>&-
  75                 {
  76                         _e1=0
  77                         bzr fast-export --plain \
  78                                 --export-marks="$(pwd)/dfe-marks" \
  79                                 --import-marks="$(pwd)/dfe-marks" "$1" 3>&- || _e1=$?
  80                         echo $_e1 >&3
  81                 } | \
  82                 {
  83                         _e2=0
  84                         git fast-import \
  85                                 --export-marks="$(pwd)/gfi-marks" \
  86                                 --export-pack-edges="$(pwd)/gfi-packs" \
  87                                 --import-marks="$(pwd)/gfi-marks" \
  88                                 --force 3>&- || _e2=$?
  89                         echo $_e2 >&3
  90                 }
  91         )
  92         EOT
  93         exec 3>&-
  94         [ "$_err1" = 0 -a "$_err2" = 0 ]
  95         return $?
  96 )
  97
  98 [ -n "$cfg_mirror" ] || { echo "Mirroring is disabled" >&2; exit 0; }
  99
 100 umask 002
 101 [ "$cfg_permission_control" != "Hooks" ] || umask 000
 102 clean_git_env
 103
 104 proj="${1%.git}"
 105 cd "$cfg_reporoot/$proj.git"
 106
 107 trap 'if [ $? != 0 ]; then echo "update failed dir: $PWD" >&2; fi; rm -f "$bang_log"' EXIT
 108 trap 'exit 130' INT
 109 trap 'exit 143' TERM
 110
 111 if check_interval lastrefresh $cfg_min_mirror_interval; then
 112         progress "= [$proj] update skip (last at $(config_get lastrefresh))"
 113         exit 0
 114 fi
 115 progress "+ [$proj] update (`date`)"
 116
 117 # Any pre-existing FETCH_HEAD from a previous fetch failed or not is garbage
 118 rm -f FETCH_HEAD
 119
 120 # A previous failed update attempt can leave a huge tmp_pack_XXXXXX file behind.
 121 # Since no pushes are allowed to mirrors, we know that any such files that exist
 122 # at this point in time are garbage and can be safely deleted, we do not even
 123 # need to check how old they are.  A tmp_idx_XXXXXX file is also created during
 124 # the later stages of the fetch process, so we kill any of those as well.
 125 find objects/pack -maxdepth 1 -type f -name "tmp_pack_?*" -print0 | xargs -0 rm -f
 126 find objects/pack -maxdepth 1 -type f -name "tmp_idx_?*" -print0 | xargs -0 rm -f
 127
 128 # Make sure we have a reflogs subdirectory and abort the update if not
 129 # This should not count as a normal "bang" failure if unsuccessful
 130 [ -d reflogs ] || mkdir -p reflogs >/dev/null 2>&1 || :
 131 [ -d reflogs ]
 132
 133 bang_setup
 134 bang_action="update"
 135 bang_trap() {
 136         if [ -n "$1" ]; then
 137                 # Throttle retries
 138                 # Since gitweb shows the .last_refresh date, it's safe to update
 139                 # gitweb.lastrefresh to throttle the updates w/o corrupting the
 140                 # last refresh date display on the gitweb summary page
 141                 # It's therefore important that we do NOT touch .last_refresh here
 142                 config_set lastrefresh "$(date "$datefmt")"
 143         fi
 144 }
 145
 146 bang echo "Project: $proj"
 147 bang echo ""
 148 mail="$(config_get owner || :)"
 149 url="$(config_get baseurl || :)"
 150 case "$url" in *"       "*|*" "*|"")
 151         bang_eval 'echo "Bad mirror URL (\"$url\")"; ! :'
 152         exit 1
 153 esac
 154 bang echo "Mirroring from URL \"$url\""
 155 bang echo ""
 156 statusok="$(git config --bool gitweb.statusupdates 2>/dev/null || echo true)"
 157 mailaddrs=''
 158 [ "$statusok" = "false" -o -z "$mail" ] || mailaddrs="$mail"
 159 [ -z "$cfg_admincc" -o "$cfg_admincc" = "0" -o -z "$cfg_admin" ] || \
 160 if [ -z "$mailaddrs" ]; then mailaddrs="$cfg_admin"; else mailaddrs="$mailaddrs,$cfg_admin"; fi
 161
 162 bang_eval "git for-each-ref --format '%(refname) %(objectname)' >.refs-temp"
 163 bang_eval "LC_ALL=C sort -b -k1,1 <.refs-temp >.refs-before"
 164 ! [ -e .delaygc ] || >.allowgc || :
 165 svnpackcreated=
 166
 167 case "$url" in
 168         svn://* | svn+http://* | svn+https://* | svn+file://* | svn+ssh://*)
 169                 [ -n "$cfg_mirror_svn" ] || { echo "Mirroring svn is disabled" >&2; exit 0; }
 170                 # Use an 'anonsvn' username as is commonly used for anonymous svn
 171                 # Use an 'anonsvn' password as is commonly used for anonymous svn
 172                 GIT_ASKPASS_PASSWORD=anonsvn
 173                 export GIT_ASKPASS_PASSWORD
 174                 # Update the git svn url to match baseurl but be cognizant of any
 175                 # needed prefix changes.  See the comments in taskd/clone.sh about
 176                 # why we need to put up with a prefix in the first place.
 177                 case "$url" in svn+ssh://*) svnurl="$url";; *) svnurl="${url#svn+}";; esac
 178                 svnurl="${svnurl%/}"
 179                 svnurlold="$(config_get svnurl || :)"
 180                 if [ "$svnurl" != "$svnurlold" ]; then
 181                         # We better already have an svn-remote.svn.fetch setting
 182                         bang test -n "$(git config --get-all svn-remote.svn.fetch || :)"
 183                         # the only way to truly know what the proper prefix is
 184                         # is to attempt a fresh git-svn init -s on the new url
 185                         rm -rf svn-new-url || :
 186                         # We require svn info to succeed on the URL otherwise it's
 187                         # simply not a valid URL and without using -s on the init it
 188                         # will not otherwise be tested until the fetch
 189                         bang eval 'svn --non-interactive --username anonsvn --password anonsvn info "$svnurl" > /dev/null'
 190                         bang mkdir svn-new-url
 191                         GIT_DIR=svn-new-url bang git init --bare --quiet
 192                         # We initially use -s for the init which will possibly shorten
 193                         # the URL.  However, the shortening can fail if a password is
 194                         # not required for the longer version but is for the shorter,
 195                         # so try again without -s if the -s version fails.
 196                         cmdstr='git svn init --username=anonsvn --prefix "" -s "$svnurl" < /dev/null > /dev/null 2>&1 || '
 197                         cmdstr="$cmdstr"'git svn init --username=anonsvn --prefix "" "$svnurl" < /dev/null > /dev/null 2>&1'
 198                         GIT_DIR=svn-new-url bang eval "$cmdstr"
 199                         gitsvnurl="$(GIT_DIR=svn-new-url git config --get svn-remote.svn.url || :)"
 200                         gitsvnfetch="$(GIT_DIR=svn-new-url git config --get svn-remote.svn.fetch || :)"
 201                         gitsvnprefixnew="${gitsvnfetch%%:*}"
 202                         gitsvnsuffixnew="${gitsvnprefixnew##*/}"
 203                         gitsvnprefixnew="${gitsvnprefixnew%$gitsvnsuffixnew}"
 204                         rm -rf svn-new-url || :
 205                         # Using GIT_DIR= with bang leaves it set to svn-new-url, so reset it to .
 206                         GIT_DIR=.
 207                         if [ "$gitsvnurl" != "$(git config --get svn-remote.svn.url || :)" ]; then
 208                                 # The url has been changed.
 209                                 # We must update the url and replace the prefix on all config items
 210                                 gitsvnfetch="$(git config --get-all svn-remote.svn.fetch | head -1 || :)"
 211                                 gitsvnprefixold="${gitsvnfetch%%:*}"
 212                                 gitsvnsuffixold="${gitsvnprefixold##*/}"
 213                                 gitsvnprefixold="${gitsvnprefixold%$gitsvnsuffixold}"
 214                                 git config --remove-section 'svn-remote.svnnew' 2>/dev/null || :
 215                                 git config 'svn-remote.svnnew.url' "$gitsvnurl"
 216                                 { git config --get-regexp '^svn-remote\.svn\.' || :; } | \
 217                                 { while read sname sval; do
 218                                         case "$sname" in
 219                                         svn-remote.svn.fetch|svn-remote.svn.branches|svn-remote.svn.tags)
 220                                                 sname="${sname#svn-remote.svn.}"
 221                                                 sval="${sval#$gitsvnprefixold}"
 222                                                 bang git config --add "svn-remote.svnnew.$sname" "${gitsvnprefixnew}$sval"
 223                                         esac
 224                                 done; }
 225                                 bang git config -f svn/.metadata svn-remote.svn.reposRoot "$gitsvnurl"
 226                                 bang git config --remove-section svn-remote.svn
 227                                 bang git config --rename-section svn-remote.svnnew svn-remote.svn
 228                         fi
 229                         bang config_set svnurl "$svnurl"
 230                 fi
 231                 # remove any stale *.lock files greater than 1 hour old in case
 232                 # git-svn was killed on the last update because it took too long
 233                 find svn -type f -name '*.lock' -mmin +60 -print0 2>/dev/null | xargs -0 rm -f
 234                 # remember the starting time so we can easily combine fetched loose objects
 235                 # we sleep for 1 second after creating .svnpack to make sure all objects are newer
 236                 if ! [ -e .svnpack ]; then
 237                         svnpackcreated=1
 238                         rm -f .svnpack
 239                         >.svnpack
 240                         sleep 1
 241                 fi
 242                 GIT_DIR=. bang git svn fetch --log-window-size=$var_log_window_size --username=anonsvn --quiet < /dev/null
 243                 # git svn does not preserve group permissions in the svn subdirectory
 244                 chmod -R ug+rw,o+r svn
 245                 # git svn also leaves behind ref turds that end with @nnn
 246                 # We get rid of them now
 247                 git for-each-ref --format='%(objectname) %(refname)' | \
 248                 { while read sha1 ref; do
 249                         case "$ref" in
 250                         ?*@[1-9]|?*@[1-9][0-9]|?*@[1-9][0-9][0-9]|?*@[1-9][0-9][0-9][0-9]|\
 251                         ?*@[1-9][0-9][0-9][0-9][0-9]|?*@[1-9][0-9][0-9][0-9][0-9][0-9]|\
 252                         ?*@[1-9][0-9][0-9][0-9][0-9][0-9][0-9]|\
 253                         ?*@[1-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9])
 254                                 git update-ref -d "$ref"
 255                         esac
 256                 done; }
 257                 unset GIT_ASKPASS_PASSWORD
 258                 ;;
 259         darcs://*)
 260                 [ -n "$cfg_mirror_darcs" ] || { echo "Mirroring darcs is disabled" >&2; exit 0; }
 261                 httpurl="http://${url#darcs://}"
 262                 # remove any stale lock files greater than 1 hour old in case
 263                 # darcs_fast_export was killed on the last update because it took too long
 264                 find *.darcs -maxdepth 2 -type f -name 'lock' -mmin +60 -print0 2>/dev/null | xargs -0 rm -f
 265                 bang git_darcs_fetch "$httpurl"
 266                 ;;
 267         bzr://*)
 268                 [ -n "$cfg_mirror_bzr" ] || { echo "Mirroring bzr is disabled" >&2; exit 0; }
 269                 bzrurl="${url#bzr://}"
 270                 bang git_bzr_fetch "$bzrurl"
 271                 ;;
 272         hg+http://* | hg+https://* | hg+file://* | hg+ssh://*)
 273                 [ -n "$cfg_mirror_hg" ] || { echo "Mirroring hg is disabled" >&2; exit 0; }
 274                 # We just remove hg+ here, so hg+http://... becomes http://...
 275                 hgurl="${url#hg+}"
 276                 # Fetch any new updates
 277                 bang hg -R "$(pwd)/repo.hg" pull
 278                 # Do the fast-export | fast-import
 279                 bang git_hg_fetch
 280                 ;;
 281         *)
 282                 [ "$url" = "$(git config --get remote.origin.url || :)" ] || bang config_set_raw remote.origin.url "$url"
 283                 pruneopt=--prune
 284                 [ "$(git config --bool fetch.prune 2>/dev/null || :)" != "false" ] || pruneopt=
 285                 git_add_config 'fetch.unpackLimit=1'
 286                 # Note the git config documentation is wrong
 287                 # transfer.unpackLimit, if set, overrides fetch.unpackLimit
 288                 git_add_config 'transfer.unpackLimit=1'
 289                 if ! is_gfi_mirror_url "$url"; then
 290                         lastwasclean=
 291                         [ "$(git config --bool girocco.lastupdateclean 2>/dev/null || :)" != "true" ] || lastwasclean=1
 292                         nextisclean=
 293                         [ "$(git config --bool girocco.cleanmirror 2>/dev/null || :)" != "true" ] || nextisclean=1
 294                         if [ "$nextisclean" != "$lastwasclean" ]; then
 295                                 if [ -n "$nextisclean" ]; then
 296                                         git config --replace-all remote.origin.fetch "+refs/heads/*:refs/heads/*"
 297                                         git config --add remote.origin.fetch "+refs/tags/*:refs/tags/*"
 298                                         git config --add remote.origin.fetch "+refs/notes/*:refs/notes/*"
 299                                 else
 300                                         git config --replace-all remote.origin.fetch "+refs/*:refs/*"
 301                                 fi
 302                         fi
 303                 fi
 304                 # remember the starting time so we can easily detect new packs for fast-import mirrors
 305                 # we sleep for 1 second after creating .gfipack to make sure all packs are newer
 306                 if is_gfi_mirror_url "$url" && [ ! -e .gfipack ]; then
 307                         rm -f .gfipack
 308                         >.gfipack
 309                         sleep 1
 310                 fi
 311                 GIT_SSL_NO_VERIFY=1 bang git remote update $pruneopt
 312                 if ! is_gfi_mirror_url "$url" && [ "$nextisclean" != "$lastwasclean" ]; then
 313                         if [ -n "$nextisclean" ]; then
 314                                 # We must manually purge the unclean refs now as even prune won't do it
 315                                 git for-each-ref --format='%(refname)' |
 316                                 while read -r aref; do
 317                                         case "$aref" in refs/heads/*|refs/tags/*|refs/notes/*) :;; *)
 318                                                 git update-ref -d "$aref"
 319                                         esac
 320                                 done
 321                         fi
 322                         git config --bool girocco.lastupdateclean ${nextisclean:-0}
 323                 fi
 324                 if [ -e .gfipack ] && is_gfi_mirror_url "$url"; then
 325                         find objects/pack -type f -newer .gfipack -name "pack-$octet20.pack" -print >>gfi-packs
 326                         rm -f .gfipack
 327                 fi
 328                 ;;
 329 esac
 330
 331 # The objects subdirectories permissions must be updated now.
 332 # In the case of a dumb http clone, the permissions will not be correct
 333 # (missing group write) despite the core.sharedrepository=1 setting!
 334 # The objects themselves seem to have the correct permissions.
 335 # This problem appears to have been fixed in the most recent git versions.
 336 perms=g+w
 337 [ "$cfg_permission_control" != "Hooks" ] || perms=go+w
 338 chmod $perms $(find objects -maxdepth 1 -type d) 2>/dev/null || :
 339
 340 bang git update-server-info
 341
 342 # We maintain the last refresh date in two places deliberately
 343 # so that it's available as part of the config data and also
 344 # as a standalone file timestamp that can be accessed without git.
 345 bang config_set lastrefresh "$(date "$datefmt")"
 346 { >.last_refresh; } 2>/dev/null || :
 347
 348 # Check to see if any refs changed
 349 bang_eval "git for-each-ref --format '%(refname) %(objectname)' >.refs-temp"
 350 bang_eval "LC_ALL=C sort -b -k1,1 <.refs-temp >.refs-after"
 351 refschanged=
 352 cmp -s .refs-before .refs-after || refschanged=1
 353
 354 # There's no way to get git svn to fetch packs, so we always need to run
 355 # a mini-gc afterwards if svn actually fetched anything, but that's okay
 356 # because it will be very quick
 357 if [ -z "$refschanged" -a -n "$svnpackcreated" ]; then
 358         # we created the .svnpack but didn't actually fetch anything
 359         # so remove it to avoid forcing a mini-gc if not necessary
 360         rm -f .svnpack
 361 fi
 362 if [ -e .svnpack ] && ! [ -e .needsgc ]; then
 363         >.needsgc
 364 fi
 365
 366 # Force a mini-gc if $Girocco::Config::delay_gfi_redelta is false and there's
 367 # at least one gfi pack present now
 368 if [ -z "$cfg_delay_gfi_redelta" ] && ! [ -e .needsgc ] && \
 369    [ -f gfi-packs -a -s gfi-packs ] && is_gfi_mirror_url "$url"; then
 370         >.needsgc
 371 fi
 372
 373 # Activate a mini-gc if there are at least 20 packs present now
 374 if ! [ -e .needsgc ]; then
 375         packs=
 376         { packs="$(list_packs --quiet --count --exclude-no-idx objects/pack || :)" || :; } 2>/dev/null
 377         if [ -n "$packs" ] && [ "$packs" -ge 20 ]; then
 378                 >.needsgc
 379         fi
 380 fi
 381
 382 # Look at which refs changed and trigger ref-change for these
 383 sockpath="$cfg_chroot/etc/taskd.socket"
 384 if [ -n "$refschanged" ]; then
 385         bang config_set lastreceive "$(date '+%a, %d %b %Y %T %z')"
 386         # We always use UTC for the log timestamp so that chroot and non-chroot match up.
 387         # We don't have to worry about multiple log files since only one update runs
 388         lognamets="$(TZ=UTC date '+%Y%m%d_%H%M%S')"
 389         loghhmmss="${lognamets##*_}"
 390         logname="reflogs/${lognamets%%_*}"
 391         # We freshen the mod time to now on any old or new ref that is a loose object
 392         # For old refs we do it so we will be able to keep them around for 1 day
 393         # For new refs we do it in case we are about to run gc and the new ref
 394         # actually points to an oldish loose object that had been unreachable
 395         # We probably do not need to do it for new refs as Git tries to do that,
 396         # but since we're already doing it for old refs (which Git does not do),
 397         # it's almost no extra work for new refs, just in case.
 398         {
 399                 echo "ref-changes %$proj% $proj"
 400                 LC_ALL=C join .refs-before .refs-after |
 401                 LC_ALL=C sed -e '/^[^ ][^ ]* \([^ ][^ ]*\) \1$/d' |
 402                         while read ref old new; do
 403                                 echo "$loghhmmss $old $new $ref" >&3
 404                                 freshen_loose_objects "$old" "$new"
 405                                 echo "$old $new $ref"
 406                         done
 407                 LC_ALL=C join -v 1 .refs-before .refs-after |
 408                         while read ref old; do
 409                                 echo "$loghhmmss $old 0000000000000000000000000000000000000000 $ref" >&3
 410                                 freshen_loose_objects "$old"
 411                                 echo "$old 0000000000000000000000000000000000000000 $ref"
 412                         done
 413                 LC_ALL=C join -v 2 .refs-before .refs-after |
 414                         while read ref new; do
 415                                 echo "$loghhmmss 0000000000000000000000000000000000000000 $new $ref" >&3
 416                                 freshen_loose_objects "$new"
 417                                 echo "0000000000000000000000000000000000000000 $new $ref"
 418                         done
 419                 git for-each-ref --format='%(objectname) %(objectname) %(refname)' refs/heads
 420         } >.refs-temp 3>>"$logname"
 421         if [ -S "$sockpath" ]; then
 422                         nc_openbsd -w 15 -U "$sockpath" <.refs-temp || :
 423         fi
 424         bang config_set lastchange "$(date '+%a, %d %b %Y %T %z')"
 425         bang_eval "git for-each-ref --sort=-committerdate --format='%(committerdate:iso8601)' \
 426                 --count=1 refs/heads > info/lastactivity"
 427         ! [ -d htmlcache ] || { >htmlcache/changed; } 2>/dev/null || :
 428         # While unlikely, it is conceivable that several ref updates have occurred that
 429         # did not actually create any packs.  In that case we could build up a large
 430         # number of log files so request gc if there are more than 50 of them now.
 431         # In the case of a mirror this is a practical impossibility but we check anyway.
 432         if ! [ -e .needsgc ]; then
 433                 logfiles=
 434                 { logfiles="$(($(find reflogs -maxdepth 1 -type f -print | wc -l || :)+0))" || :; } 2>/dev/null
 435                 if [ -n "$logfiles" ] && [ "$logfiles" -ge 50 ]; then
 436                         >.needsgc
 437                 fi
 438         fi
 439         rm -f .delaygc .allowgc
 440 fi
 441
 442 # If the repository does not yet have a valid HEAD symref try to set one
 443 # If an empty repository was cloned and then later becomes unempty you just
 444 # lose out on the fancy "symref=HEAD:" logic and get this version instead
 445 check_and_set_head || :
 446
 447 rm -f .refs-before .refs-after .refs-temp FETCH_HEAD
 448
 449 if is_banged; then
 450         [ -z "$mailaddrs" ] || ! was_banged_message_sent ||
 451         {
 452                 echo "$proj update succeeded - failure recovery"
 453                 echo "this status message may be disabled on the project admin page"
 454         } | mail -s "[$cfg_name] $proj update succeeded" "$mailaddrs" || :
 455         bang_reset
 456 fi
 457
 458 progress "- [$proj] update (`date`)"