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_action="update"
 129 bang_trap() {
 130         if [ -n "$1" ]; then
 131                 # Throttle retries
 132                 # Since gitweb shows the .last_refresh date, it's safe to update
 133                 # gitweb.lastrefresh to throttle the updates w/o corrupting the
 134                 # last refresh date display on the gitweb summary page
 135                 # It's therefore important that we do NOT touch .last_refresh here
 136                 config_set lastrefresh "$(date "$datefmt")"
 137         fi
 138 }
 139
 140 bang echo "Project: $proj"
 141 bang echo ""
 142 mail="$(config_get owner || :)"
 143 url="$(config_get baseurl || :)"
 144 case "$url" in *"       "*|*" "*|"")
 145         bang_eval 'echo "Bad mirror URL (\"$url\")"; ! :'
 146         exit 1
 147 esac
 148 bang echo "Mirroring from URL \"$url\""
 149 bang echo ""
 150 statusok="$(git config --bool gitweb.statusupdates 2>/dev/null || echo true)"
 151 mailaddrs=''
 152 [ "$statusok" = "false" -o -z "$mail" ] || mailaddrs="$mail"
 153 [ -z "$cfg_admincc" -o "$cfg_admincc" = "0" -o -z "$cfg_admin" ] || \
 154 if [ -z "$mailaddrs" ]; then mailaddrs="$cfg_admin"; else mailaddrs="$mailaddrs,$cfg_admin"; fi
 155
 156 bang_eval "git for-each-ref --format '%(refname) %(objectname)' >.refs-temp"
 157 bang_eval "LC_ALL=C sort -b -k1,1 <.refs-temp >.refs-before"
 158 ! [ -e .delaygc ] || >.allowgc || :
 159 svnpackcreated=
 160
 161 case "$url" in
 162         svn://* | svn+http://* | svn+https://* | svn+file://* | svn+ssh://*)
 163                 [ -n "$cfg_mirror_svn" ] || { echo "Mirroring svn is disabled" >&2; exit 0; }
 164                 # Use an 'anonsvn' username as is commonly used for anonymous svn
 165                 # Use an 'anonsvn' password as is commonly used for anonymous svn
 166                 GIT_ASKPASS_PASSWORD=anonsvn
 167                 export GIT_ASKPASS_PASSWORD
 168                 # Update the git svn url to match baseurl but be cognizant of any
 169                 # needed prefix changes.  See the comments in taskd/clone.sh about
 170                 # why we need to put up with a prefix in the first place.
 171                 case "$url" in svn+ssh://*) svnurl="$url";; *) svnurl="${url#svn+}";; esac
 172                 svnurl="${svnurl%/}"
 173                 svnurlold="$(config_get svnurl || :)"
 174                 if [ "$svnurl" != "$svnurlold" ]; then
 175                         # We better already have an svn-remote.svn.fetch setting
 176                         bang test -n "$(git config --get-all svn-remote.svn.fetch || :)"
 177                         # the only way to truly know what the proper prefix is
 178                         # is to attempt a fresh git-svn init -s on the new url
 179                         rm -rf svn-new-url || :
 180                         # We require svn info to succeed on the URL otherwise it's
 181                         # simply not a valid URL and without using -s on the init it
 182                         # will not otherwise be tested until the fetch
 183                         bang eval 'svn --non-interactive --username anonsvn --password anonsvn info "$svnurl" > /dev/null'
 184                         bang mkdir svn-new-url
 185                         GIT_DIR=svn-new-url bang git init --bare --quiet
 186                         # We initially use -s for the init which will possibly shorten
 187                         # the URL.  However, the shortening can fail if a password is
 188                         # not required for the longer version but is for the shorter,
 189                         # so try again without -s if the -s version fails.
 190                         cmdstr='git svn init --username=anonsvn --prefix "" -s "$svnurl" < /dev/null > /dev/null 2>&1 || '
 191                         cmdstr="$cmdstr"'git svn init --username=anonsvn --prefix "" "$svnurl" < /dev/null > /dev/null 2>&1'
 192                         GIT_DIR=svn-new-url bang eval "$cmdstr"
 193                         gitsvnurl="$(GIT_DIR=svn-new-url git config --get svn-remote.svn.url || :)"
 194                         gitsvnfetch="$(GIT_DIR=svn-new-url git config --get svn-remote.svn.fetch || :)"
 195                         gitsvnprefixnew="${gitsvnfetch%%:*}"
 196                         gitsvnsuffixnew="${gitsvnprefixnew##*/}"
 197                         gitsvnprefixnew="${gitsvnprefixnew%$gitsvnsuffixnew}"
 198                         rm -rf svn-new-url || :
 199                         # Using GIT_DIR= with bang leaves it set to svn-new-url, so reset it to .
 200                         GIT_DIR=.
 201                         if [ "$gitsvnurl" != "$(git config --get svn-remote.svn.url || :)" ]; then
 202                                 # The url has been changed.
 203                                 # We must update the url and replace the prefix on all config items
 204                                 gitsvnfetch="$(git config --get-all svn-remote.svn.fetch | head -1 || :)"
 205                                 gitsvnprefixold="${gitsvnfetch%%:*}"
 206                                 gitsvnsuffixold="${gitsvnprefixold##*/}"
 207                                 gitsvnprefixold="${gitsvnprefixold%$gitsvnsuffixold}"
 208                                 git config --remove-section 'svn-remote.svnnew' 2>/dev/null || :
 209                                 git config 'svn-remote.svnnew.url' "$gitsvnurl"
 210                                 { git config --get-regexp '^svn-remote\.svn\.' || :; } | \
 211                                 { while read sname sval; do
 212                                         case "$sname" in
 213                                         svn-remote.svn.fetch|svn-remote.svn.branches|svn-remote.svn.tags)
 214                                                 sname="${sname#svn-remote.svn.}"
 215                                                 sval="${sval#$gitsvnprefixold}"
 216                                                 bang git config --add "svn-remote.svnnew.$sname" "${gitsvnprefixnew}$sval"
 217                                         esac
 218                                 done; }
 219                                 bang git config -f svn/.metadata svn-remote.svn.reposRoot "$gitsvnurl"
 220                                 bang git config --remove-section svn-remote.svn
 221                                 bang git config --rename-section svn-remote.svnnew svn-remote.svn
 222                         fi
 223                         bang config_set svnurl "$svnurl"
 224                 fi
 225                 # remove any stale *.lock files greater than 1 hour old in case
 226                 # git-svn was killed on the last update because it took too long
 227                 find svn -type f -name '*.lock' -mmin +60 -print0 2>/dev/null | xargs -0 rm -f
 228                 # remember the starting time so we can easily combine fetched loose objects
 229                 # we sleep for 1 second after creating .svnpack to make sure all objects are newer
 230                 if ! [ -e .svnpack ]; then
 231                         svnpackcreated=1
 232                         rm -f .svnpack
 233                         >.svnpack
 234                         sleep 1
 235                 fi
 236                 GIT_DIR=. bang git svn fetch --log-window-size=$var_log_window_size --username=anonsvn --quiet < /dev/null
 237                 # git svn does not preserve group permissions in the svn subdirectory
 238                 chmod -R ug+rw,o+r svn
 239                 # git svn also leaves behind ref turds that end with @nnn
 240                 # We get rid of them now
 241                 git for-each-ref --format='%(objectname) %(refname)' | \
 242                 { while read sha1 ref; do
 243                         case "$ref" in
 244                         ?*@[1-9]|?*@[1-9][0-9]|?*@[1-9][0-9][0-9]|?*@[1-9][0-9][0-9][0-9]|\
 245                         ?*@[1-9][0-9][0-9][0-9][0-9]|?*@[1-9][0-9][0-9][0-9][0-9][0-9]|\
 246                         ?*@[1-9][0-9][0-9][0-9][0-9][0-9][0-9]|\
 247                         ?*@[1-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9])
 248                                 git update-ref -d "$ref"
 249                         esac
 250                 done; }
 251                 unset GIT_ASKPASS_PASSWORD
 252                 ;;
 253         darcs://*)
 254                 [ -n "$cfg_mirror_darcs" ] || { echo "Mirroring darcs is disabled" >&2; exit 0; }
 255                 httpurl="http://${url#darcs://}"
 256                 # remove any stale lock files greater than 1 hour old in case
 257                 # darcs_fast_export was killed on the last update because it took too long
 258                 find *.darcs -maxdepth 2 -type f -name 'lock' -mmin +60 -print0 2>/dev/null | xargs -0 rm -f
 259                 bang git_darcs_fetch "$httpurl"
 260                 ;;
 261         bzr://*)
 262                 [ -n "$cfg_mirror_bzr" ] || { echo "Mirroring bzr is disabled" >&2; exit 0; }
 263                 bzrurl="${url#bzr://}"
 264                 bang git_bzr_fetch "$bzrurl"
 265                 ;;
 266         hg+http://* | hg+https://* | hg+file://* | hg+ssh://*)
 267                 [ -n "$cfg_mirror_hg" ] || { echo "Mirroring hg is disabled" >&2; exit 0; }
 268                 # We just remove hg+ here, so hg+http://... becomes http://...
 269                 hgurl="${url#hg+}"
 270                 # Fetch any new updates
 271                 bang hg -R "$(pwd)/repo.hg" pull
 272                 # Do the fast-export | fast-import
 273                 bang git_hg_fetch
 274                 ;;
 275         *)
 276                 [ "$url" = "$(git config --get remote.origin.url || :)" ] || bang config_set_raw remote.origin.url "$url"
 277                 pruneopt=--prune
 278                 [ "$(git config --bool fetch.prune 2>/dev/null || :)" != "false" ] || pruneopt=
 279                 git_add_config 'fetch.unpackLimit=1'
 280                 # Note the git config documentation is wrong
 281                 # transfer.unpackLimit, if set, overrides fetch.unpackLimit
 282                 git_add_config 'transfer.unpackLimit=1'
 283                 if ! is_gfi_mirror_url "$url"; then
 284                         lastwasclean=
 285                         [ "$(git config --bool girocco.lastupdateclean 2>/dev/null || :)" != "true" ] || lastwasclean=1
 286                         nextisclean=
 287                         [ "$(git config --bool girocco.cleanmirror 2>/dev/null || :)" != "true" ] || nextisclean=1
 288                         if [ "$nextisclean" != "$lastwasclean" ]; then
 289                                 if [ -n "$nextisclean" ]; then
 290                                         git config --replace-all remote.origin.fetch "+refs/heads/*:refs/heads/*"
 291                                         git config --add remote.origin.fetch "+refs/tags/*:refs/tags/*"
 292                                         git config --add remote.origin.fetch "+refs/notes/*:refs/notes/*"
 293                                 else
 294                                         git config --replace-all remote.origin.fetch "+refs/*:refs/*"
 295                                 fi
 296                         fi
 297                 fi
 298                 # remember the starting time so we can easily detect new packs for fast-import mirrors
 299                 # we sleep for 1 second after creating .gfipack to make sure all packs are newer
 300                 if is_gfi_mirror_url "$url" && [ ! -e .gfipack ]; then
 301                         rm -f .gfipack
 302                         >.gfipack
 303                         sleep 1
 304                 fi
 305                 GIT_SSL_NO_VERIFY=1 bang git remote update $pruneopt
 306                 if ! is_gfi_mirror_url "$url" && [ "$nextisclean" != "$lastwasclean" ]; then
 307                         if [ -n "$nextisclean" ]; then
 308                                 # We must manually purge the unclean refs now as even prune won't do it
 309                                 git for-each-ref --format='%(refname)' |
 310                                 while read -r aref; do
 311                                         case "$aref" in refs/heads/*|refs/tags/*|refs/notes/*) :;; *)
 312                                                 git update-ref -d "$aref"
 313                                         esac
 314                                 done
 315                         fi
 316                         git config --bool girocco.lastupdateclean ${nextisclean:-0}
 317                 fi
 318                 if [ -e .gfipack ] && is_gfi_mirror_url "$url"; then
 319                         find objects/pack -type f -newer .gfipack -name "pack-$octet20.pack" -print >>gfi-packs
 320                         rm -f .gfipack
 321                 fi
 322                 ;;
 323 esac
 324
 325 # The objects subdirectories permissions must be updated now.
 326 # In the case of a dumb http clone, the permissions will not be correct
 327 # (missing group write) despite the core.sharedrepository=1 setting!
 328 # The objects themselves seem to have the correct permissions.
 329 # This problem appears to have been fixed in the most recent git versions.
 330 perms=g+w
 331 [ "$cfg_permission_control" != "Hooks" ] || perms=go+w
 332 chmod $perms $(find objects -maxdepth 1 -type d) 2>/dev/null || :
 333
 334 bang git update-server-info
 335
 336 # We maintain the last refresh date in two places deliberately
 337 # so that it's available as part of the config data and also
 338 # as a standalone file timestamp that can be accessed without git.
 339 bang config_set lastrefresh "$(date "$datefmt")"
 340 { >.last_refresh; } 2>/dev/null || :
 341
 342 # Check to see if any refs changed
 343 bang_eval "git for-each-ref --format '%(refname) %(objectname)' >.refs-temp"
 344 bang_eval "LC_ALL=C sort -b -k1,1 <.refs-temp >.refs-after"
 345 refschanged=
 346 cmp -s .refs-before .refs-after || refschanged=1
 347
 348 # There's no way to get git svn to fetch packs, so we always need to run
 349 # a mini-gc afterwards if svn actually fetched anything, but that's okay
 350 # because it will be very quick
 351 if [ -z "$refschanged" -a -n "$svnpackcreated" ]; then
 352         # we created the .svnpack but didn't actually fetch anything
 353         # so remove it to avoid forcing a mini-gc if not necessary
 354         rm -f .svnpack
 355 fi
 356 if [ -e .svnpack ] && ! [ -e .needsgc ]; then
 357         >.needsgc
 358 fi
 359
 360 # Force a mini-gc if $Girocco::Config::delay_gfi_redelta is false and there's
 361 # at least one gfi pack present now
 362 if [ -z "$cfg_delay_gfi_redelta" ] && ! [ -e .needsgc ] && \
 363    [ -f gfi-packs -a -s gfi-packs ] && is_gfi_mirror_url "$url"; then
 364         >.needsgc
 365 fi
 366
 367 # Activate a mini-gc if there are at least 20 packs present now
 368 if ! [ -e .needsgc ]; then
 369         packs=
 370         { packs="$(list_packs --quiet --count --exclude-no-idx objects/pack || :)" || :; } 2>/dev/null
 371         if [ -n "$packs" ] && [ "$packs" -ge 20 ]; then
 372                 >.needsgc
 373         fi
 374 fi
 375
 376 # Look at which refs changed and trigger ref-change for these
 377 sockpath="$cfg_chroot/etc/taskd.socket"
 378 if [ -n "$refschanged" ]; then
 379         bang config_set lastreceive "$(date '+%a, %d %b %Y %T %z')"
 380         # We freshen the mod time to now on any old or new ref that is a loose object
 381         # For old refs we do it so we will be able to keep them around for 1 day
 382         # For new refs we do it in case we are about to run gc and the new ref
 383         # actually points to an oldish loose object that had been unreachable
 384         # We probably do not need to do it for new refs as Git tries to do that,
 385         # but since we're already doing it for old refs (which Git does not do),
 386         # it's almost no extra work for new refs, just in case.
 387         {
 388                 echo "ref-changes %$proj% $proj"
 389                 LC_ALL=C join .refs-before .refs-after |
 390                 LC_ALL=C sed -e '/^[^ ][^ ]* \([^ ][^ ]*\) \1$/d' |
 391                         while read ref old new; do
 392                                 freshen_loose_objects "$old" "$new"
 393                                 echo "$old $new $ref"
 394                         done
 395                 LC_ALL=C join -v 1 .refs-before .refs-after |
 396                         while read ref old; do
 397                                 freshen_loose_objects "$old"
 398                                 echo "$old 0000000000000000000000000000000000000000 $ref"
 399                         done
 400                 LC_ALL=C join -v 2 .refs-before .refs-after |
 401                         while read ref new; do
 402                                 freshen_loose_objects "$new"
 403                                 echo "0000000000000000000000000000000000000000 $new $ref"
 404                         done
 405                 git for-each-ref --format='%(objectname) %(objectname) %(refname)' refs/heads
 406         } >.refs-temp
 407         if [ -S "$sockpath" ]; then
 408                         nc_openbsd -w 15 -U "$sockpath" <.refs-temp || :
 409         fi
 410         bang config_set lastchange "$(date '+%a, %d %b %Y %T %z')"
 411         bang_eval "git for-each-ref --sort=-committerdate --format='%(committerdate:iso8601)' \
 412                 --count=1 refs/heads > info/lastactivity"
 413         ! [ -d htmlcache ] || { >htmlcache/changed; } 2>/dev/null || :
 414         rm -f .delaygc .allowgc
 415 fi
 416
 417 # If the repository does not yet have a valid HEAD symref try to set one
 418 # If an empty repository was cloned and then later becomes unempty you just
 419 # lose out on the fancy "symref=HEAD:" logic and get this version instead
 420 check_and_set_head || :
 421
 422 rm -f .refs-before .refs-after .refs-temp FETCH_HEAD
 423
 424 if is_banged; then
 425         [ -z "$mailaddrs" ] || ! was_banged_message_sent ||
 426         {
 427                 echo "$proj update succeeded - failure recovery"
 428                 echo "this status message may be disabled on the project admin page"
 429         } | mail -s "[$cfg_name] $proj update succeeded" "$mailaddrs" || :
 430         bang_reset
 431 fi
 432
 433 progress "- [$proj] update (`date`)"