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 [ "$cfg_permission_control" != "Hooks" ] || umask 000
 102
 103 proj="${1%.git}"
 104 cd "$cfg_reporoot/$proj.git"
 105
 106 trap 'if [ $? != 0 ]; then echo "update failed dir: $PWD" >&2; fi; rm -f "$bang_log"' EXIT
 107 trap 'exit 130' INT
 108 trap 'exit 143' TERM
 109
 110 if check_interval lastrefresh $cfg_min_mirror_interval; then
 111         progress "= [$proj] update skip (last at $(config_get lastrefresh))"
 112         exit 0
 113 fi
 114 progress "+ [$proj] update (`date`)"
 115
 116 # Any pre-existing FETCH_HEAD from a previous fetch failed or not is garbage
 117 rm -f FETCH_HEAD
 118
 119 # A previous failed update attempt can leave a huge tmp_pack_XXXXXX file behind.
 120 # Since no pushes are allowed to mirrors, we know that any such files that exist
 121 # at this point in time are garbage and can be safely deleted, we do not even
 122 # need to check how old they are.  A tmp_idx_XXXXXX file is also created during
 123 # the later stages of the fetch process, so we kill any of those as well.
 124 find objects/pack -maxdepth 1 -type f -name "tmp_pack_?*" -print0 | xargs -0 rm -f
 125 find objects/pack -maxdepth 1 -type f -name "tmp_idx_?*" -print0 | xargs -0 rm -f
 126
 127 bang_setup
 128 bang_once=1
 129 bang_action="update"
 130
 131 bang echo "Project: $proj"
 132 bang echo ""
 133 mail="$(config_get owner || :)"
 134 url="$(config_get baseurl || :)"
 135 case "$url" in *"       "*|*" "*|"")
 136         bang_eval 'echo "Bad mirror URL (\"$url\")"; ! :'
 137         exit 1
 138 esac
 139 bang echo "Mirroring from URL \"$url\""
 140 bang echo ""
 141 statusok="$(git config --bool gitweb.statusupdates 2>/dev/null || echo true)"
 142 mailaddrs=''
 143 [ "$statusok" = "false" -o -z "$mail" ] || mailaddrs="$mail"
 144 [ -z "$cfg_admincc" -o "$cfg_admincc" = "0" -o -z "$cfg_admin" ] || \
 145 if [ -z "$mailaddrs" ]; then mailaddrs="$cfg_admin"; else mailaddrs="$mailaddrs,$cfg_admin"; fi
 146
 147 bang_eval "git for-each-ref --format '%(refname) %(objectname)' >.refs-temp"
 148 bang_eval "LC_ALL=C sort -b -k1,1 <.refs-temp >.refs-before"
 149 ! [ -e .delaygc ] || >.allowgc || :
 150 svnpackcreated=
 151
 152 case "$url" in
 153         svn://* | svn+http://* | svn+https://* | svn+file://* | svn+ssh://*)
 154                 [ -n "$cfg_mirror_svn" ] || { echo "Mirroring svn is disabled" >&2; exit 0; }
 155                 # Use an 'anonsvn' username as is commonly used for anonymous svn
 156                 # Use an 'anonsvn' password as is commonly used for anonymous svn
 157                 GIT_ASKPASS_PASSWORD=anonsvn
 158                 export GIT_ASKPASS_PASSWORD
 159                 # Update the git svn url to match baseurl but be cognizant of any
 160                 # needed prefix changes.  See the comments in taskd/clone.sh about
 161                 # why we need to put up with a prefix in the first place.
 162                 case "$url" in svn+ssh://*) svnurl="$url";; *) svnurl="${url#svn+}";; esac
 163                 svnurl="${svnurl%/}"
 164                 svnurlold="$(config_get svnurl || :)"
 165                 if [ "$svnurl" != "$svnurlold" ]; then
 166                         # We better already have an svn-remote.svn.fetch setting
 167                         bang test -n "$(git config --get-all svn-remote.svn.fetch || :)"
 168                         # the only way to truly know what the proper prefix is
 169                         # is to attempt a fresh git-svn init -s on the new url
 170                         rm -rf svn-new-url || :
 171                         # We require svn info to succeed on the URL otherwise it's
 172                         # simply not a valid URL and without using -s on the init it
 173                         # will not otherwise be tested until the fetch
 174                         bang eval 'svn --non-interactive --username anonsvn --password anonsvn info "$svnurl" > /dev/null'
 175                         bang mkdir svn-new-url
 176                         GIT_DIR=svn-new-url bang git init --bare --quiet
 177                         # We initially use -s for the init which will possibly shorten
 178                         # the URL.  However, the shortening can fail if a password is
 179                         # not required for the longer version but is for the shorter,
 180                         # so try again without -s if the -s version fails.
 181                         cmdstr='git svn init --username=anonsvn --prefix "" -s "$svnurl" < /dev/null > /dev/null 2>&1 || '
 182                         cmdstr="$cmdstr"'git svn init --username=anonsvn --prefix "" "$svnurl" < /dev/null > /dev/null 2>&1'
 183                         GIT_DIR=svn-new-url bang eval "$cmdstr"
 184                         gitsvnurl="$(GIT_DIR=svn-new-url git config --get svn-remote.svn.url || :)"
 185                         gitsvnfetch="$(GIT_DIR=svn-new-url git config --get svn-remote.svn.fetch || :)"
 186                         gitsvnprefixnew="${gitsvnfetch%%:*}"
 187                         gitsvnsuffixnew="${gitsvnprefixnew##*/}"
 188                         gitsvnprefixnew="${gitsvnprefixnew%$gitsvnsuffixnew}"
 189                         rm -rf svn-new-url || :
 190                         # Using GIT_DIR= with bang leaves it set to svn-new-url, so reset it to .
 191                         GIT_DIR=.
 192                         if [ "$gitsvnurl" != "$(git config --get svn-remote.svn.url || :)" ]; then
 193                                 # The url has been changed.
 194                                 # We must update the url and replace the prefix on all config items
 195                                 gitsvnfetch="$(git config --get-all svn-remote.svn.fetch | head -1 || :)"
 196                                 gitsvnprefixold="${gitsvnfetch%%:*}"
 197                                 gitsvnsuffixold="${gitsvnprefixold##*/}"
 198                                 gitsvnprefixold="${gitsvnprefixold%$gitsvnsuffixold}"
 199                                 git config --remove-section 'svn-remote.svnnew' 2>/dev/null || :
 200                                 git config 'svn-remote.svnnew.url' "$gitsvnurl"
 201                                 { git config --get-regexp '^svn-remote\.svn\.' || :; } | \
 202                                 { while read sname sval; do
 203                                         case "$sname" in
 204                                         svn-remote.svn.fetch|svn-remote.svn.branches|svn-remote.svn.tags)
 205                                                 sname="${sname#svn-remote.svn.}"
 206                                                 sval="${sval#$gitsvnprefixold}"
 207                                                 bang git config --add "svn-remote.svnnew.$sname" "${gitsvnprefixnew}$sval"
 208                                         esac
 209                                 done; }
 210                                 bang git config -f svn/.metadata svn-remote.svn.reposRoot "$gitsvnurl"
 211                                 bang git config --remove-section svn-remote.svn
 212                                 bang git config --rename-section svn-remote.svnnew svn-remote.svn
 213                         fi
 214                         bang config_set svnurl "$svnurl"
 215                 fi
 216                 # remove any stale *.lock files greater than 1 hour old in case
 217                 # git-svn was killed on the last update because it took too long
 218                 find svn -type f -name '*.lock' -mmin +60 -print0 2>/dev/null | xargs -0 rm -f
 219                 # remember the starting time so we can easily combine fetched loose objects
 220                 # we sleep for 1 second after creating .svnpack to make sure all objects are newer
 221                 if ! [ -e .svnpack ]; then
 222                         svnpackcreated=1
 223                         rm -f .svnpack
 224                         >.svnpack
 225                         sleep 1
 226                 fi
 227                 GIT_DIR=. bang git svn fetch --log-window-size=$var_log_window_size --username=anonsvn --quiet < /dev/null
 228                 # git svn does not preserve group permissions in the svn subdirectory
 229                 chmod -R ug+rw,o+r svn
 230                 # git svn also leaves behind ref turds that end with @nnn
 231                 # We get rid of them now
 232                 git for-each-ref --format='%(objectname) %(refname)' | \
 233                 { while read sha1 ref; do
 234                         case "$ref" in
 235                         ?*@[1-9]|?*@[1-9][0-9]|?*@[1-9][0-9][0-9]|?*@[1-9][0-9][0-9][0-9]|\
 236                         ?*@[1-9][0-9][0-9][0-9][0-9]|?*@[1-9][0-9][0-9][0-9][0-9][0-9]|\
 237                         ?*@[1-9][0-9][0-9][0-9][0-9][0-9][0-9]|\
 238                         ?*@[1-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9])
 239                                 git update-ref -d "$ref"
 240                         esac
 241                 done; }
 242                 unset GIT_ASKPASS_PASSWORD
 243                 ;;
 244         darcs://*)
 245                 [ -n "$cfg_mirror_darcs" ] || { echo "Mirroring darcs is disabled" >&2; exit 0; }
 246                 httpurl="http://${url#darcs://}"
 247                 # remove any stale lock files greater than 1 hour old in case
 248                 # darcs_fast_export was killed on the last update because it took too long
 249                 find *.darcs -maxdepth 2 -type f -name 'lock' -mmin +60 -print0 2>/dev/null | xargs -0 rm -f
 250                 bang git_darcs_fetch "$httpurl"
 251                 ;;
 252         bzr://*)
 253                 [ -n "$cfg_mirror_bzr" ] || { echo "Mirroring bzr is disabled" >&2; exit 0; }
 254                 bzrurl="${url#bzr://}"
 255                 bang git_bzr_fetch "$bzrurl"
 256                 ;;
 257         hg+http://* | hg+https://* | hg+file://* | hg+ssh://*)
 258                 [ -n "$cfg_mirror_hg" ] || { echo "Mirroring hg is disabled" >&2; exit 0; }
 259                 # We just remove hg+ here, so hg+http://... becomes http://...
 260                 hgurl="${url#hg+}"
 261                 # Fetch any new updates
 262                 bang hg -R "$(pwd)/repo.hg" pull
 263                 # Do the fast-export | fast-import
 264                 bang git_hg_fetch
 265                 ;;
 266         *)
 267                 [ "$url" = "$(git config --get remote.origin.url || :)" ] || bang config_set_raw remote.origin.url "$url"
 268                 pruneopt=--prune
 269                 [ "$(git config --bool fetch.prune 2>/dev/null || :)" != "false" ] || pruneopt=
 270                 git_add_config 'fetch.unpackLimit=1'
 271                 # Note the git config documentation is wrong
 272                 # transfer.unpackLimit, if set, overrides fetch.unpackLimit
 273                 git_add_config 'transfer.unpackLimit=1'
 274                 if ! is_gfi_mirror_url "$url"; then
 275                         lastwasclean=
 276                         [ "$(git config --bool girocco.lastupdateclean 2>/dev/null || :)" != "true" ] || lastwasclean=1
 277                         nextisclean=
 278                         [ "$(git config --bool girocco.cleanmirror 2>/dev/null || :)" != "true" ] || nextisclean=1
 279                         if [ "$nextisclean" != "$lastwasclean" ]; then
 280                                 if [ -n "$nextisclean" ]; then
 281                                         git config --replace-all remote.origin.fetch "+refs/heads/*:refs/heads/*"
 282                                         git config --add remote.origin.fetch "+refs/tags/*:refs/tags/*"
 283                                         git config --add remote.origin.fetch "+refs/notes/*:refs/notes/*"
 284                                 else
 285                                         git config --replace-all remote.origin.fetch "+refs/*:refs/*"
 286                                 fi
 287                         fi
 288                 fi
 289                 # remember the starting time so we can easily detect new packs for fast-import mirrors
 290                 # we sleep for 1 second after creating .gfipack to make sure all packs are newer
 291                 if is_gfi_mirror_url "$url" && [ ! -e .gfipack ]; then
 292                         rm -f .gfipack
 293                         >.gfipack
 294                         sleep 1
 295                 fi
 296                 GIT_SSL_NO_VERIFY=1 bang git remote update $pruneopt
 297                 if ! is_gfi_mirror_url "$url" && [ "$nextisclean" != "$lastwasclean" ]; then
 298                         if [ -n "$nextisclean" ]; then
 299                                 # We must manually purge the unclean refs now as even prune won't do it
 300                                 git for-each-ref --format='%(refname)' |
 301                                 while read -r aref; do
 302                                         case "$aref" in refs/heads/*|refs/tags/*|refs/notes/*) :;; *)
 303                                                 git update-ref -d "$aref"
 304                                         esac
 305                                 done
 306                         fi
 307                         git config --bool girocco.lastupdateclean ${nextisclean:-0}
 308                 fi
 309                 if [ -e .gfipack ] && is_gfi_mirror_url "$url"; then
 310                         find objects/pack -type f -newer .gfipack -name "pack-$octet20.pack" -print >>gfi-packs
 311                         rm -f .gfipack
 312                 fi
 313                 ;;
 314 esac
 315
 316 # The objects subdirectories permissions must be updated now.
 317 # In the case of a dumb http clone, the permissions will not be correct
 318 # (missing group write) despite the core.sharedrepository=1 setting!
 319 # The objects themselves seem to have the correct permissions.
 320 # This problem appears to have been fixed in the most recent git versions.
 321 perms=g+w
 322 [ "$cfg_permission_control" != "Hooks" ] || perms=go+w
 323 chmod $perms $(find objects -maxdepth 1 -type d) 2>/dev/null || :
 324
 325 bang git update-server-info
 326
 327 # We maintain the last refresh date in two places deliberately
 328 # so that it's available as part of the config data and also
 329 # as a standalone file timestamp that can be accessed without git.
 330 bang config_set lastrefresh "$(date "$datefmt")"
 331 { >.last_refresh; } 2>/dev/null || :
 332
 333 # Check to see if any refs changed
 334 bang_eval "git for-each-ref --format '%(refname) %(objectname)' >.refs-temp"
 335 bang_eval "LC_ALL=C sort -b -k1,1 <.refs-temp >.refs-after"
 336 refschanged=
 337 cmp -s .refs-before .refs-after || refschanged=1
 338
 339 # There's no way to get git svn to fetch packs, so we always need to run
 340 # a mini-gc afterwards if svn actually fetched anything, but that's okay
 341 # because it will be very quick
 342 if [ -z "$refschanged" -a -n "$svnpackcreated" ]; then
 343         # we created the .svnpack but didn't actually fetch anything
 344         # so remove it to avoid forcing a mini-gc if not necessary
 345         rm -f .svnpack
 346 fi
 347 if [ -e .svnpack ] && ! [ -e .needsgc ]; then
 348         >.needsgc
 349 fi
 350
 351 # Force a mini-gc if $Girocco::Config::delay_gfi_redelta is false and there's
 352 # at least one gfi pack present now
 353 if [ -z "$cfg_delay_gfi_redelta" ] && ! [ -e .needsgc ] && \
 354    [ -f gfi-packs -a -s gfi-packs ] && is_gfi_mirror_url "$url"; then
 355         >.needsgc
 356 fi
 357
 358 # Activate a mini-gc if there are at least 20 packs present now
 359 if ! [ -e .needsgc ]; then
 360         packs=
 361         { packs="$(list_packs --quiet --count --exclude-no-idx objects/pack || :)" || :; } 2>/dev/null
 362         if [ -n "$packs" ] && [ "$packs" -ge 20 ]; then
 363                 >.needsgc
 364         fi
 365 fi
 366
 367 # Look at which refs changed and trigger ref-change for these
 368 sockpath="$cfg_chroot/etc/taskd.socket"
 369 if [ -n "$refschanged" ]; then
 370         bang config_set lastreceive "$(date '+%a, %d %b %Y %T %z')"
 371         # We freshen the mod time to now on any old or new ref that is a loose object
 372         # For old refs we do it so we will be able to keep them around for 1 day
 373         # For new refs we do it in case we are about to run gc and the new ref
 374         # actually points to an oldish loose object that had been unreachable
 375         # We probably do not need to do it for new refs as Git tries to do that,
 376         # but since we're already doing it for old refs (which Git does not do),
 377         # it's almost no extra work for new refs, just in case.
 378         {
 379                 echo "ref-changes %$proj% $proj"
 380                 LC_ALL=C join .refs-before .refs-after |
 381                 LC_ALL=C sed -e '/^[^ ][^ ]* \([^ ][^ ]*\) \1$/d' |
 382                         while read ref old new; do
 383                                 freshen_loose_objects "$old" "$new"
 384                                 echo "$old $new $ref"
 385                         done
 386                 LC_ALL=C join -v 1 .refs-before .refs-after |
 387                         while read ref old; do
 388                                 freshen_loose_objects "$old"
 389                                 echo "$old 0000000000000000000000000000000000000000 $ref"
 390                         done
 391                 LC_ALL=C join -v 2 .refs-before .refs-after |
 392                         while read ref new; do
 393                                 freshen_loose_objects "$new"
 394                                 echo "0000000000000000000000000000000000000000 $new $ref"
 395                         done
 396                 git for-each-ref --format='%(objectname) %(objectname) %(refname)' refs/heads
 397         } >.refs-temp
 398         if [ -S "$sockpath" ]; then
 399                         nc_openbsd -w 15 -U "$sockpath" <.refs-temp || :
 400         fi
 401         bang config_set lastchange "$(date '+%a, %d %b %Y %T %z')"
 402         bang_eval "git for-each-ref --sort=-committerdate --format='%(committerdate:iso8601)' \
 403                 --count=1 refs/heads > info/lastactivity"
 404         ! [ -d htmlcache ] || { >htmlcache/changed; } 2>/dev/null || :
 405         rm -f .delaygc .allowgc
 406 fi
 407
 408 rm -f .refs-before .refs-after .refs-temp FETCH_HEAD
 409
 410 if is_banged; then
 411         [ -z "$mailaddrs" ] ||
 412         {
 413                 echo "$proj update succeeded - failure recovery"
 414                 echo "this status message may be disabled on the project admin page"
 415         } | mail -s "[$cfg_name] $proj update succeeded" "$mailaddrs" || :
 416         bang_reset
 417 fi
 418
 419 progress "- [$proj] update (`date`)"