[Mediawiki-l] Moving from one server to another

Thu Aug 18 19:25:18 UTC 2005

On Thu, Jul 21, 2005 at 05:19:24PM +0200, Brion Vibber wrote:
> Old URLs still in the parser cache can be cleared by:
> 
>   DELETE * FROM objectcache;
> 
> to clear the objectcache table in MySQL.
> 
> Note that the browser cache may still contain bogus data. You can use
> ?action=purge on indidivual pages to update their cache dates. Also
> there is a global $wgCacheEpoch variable which can be set in
> LocalSettings.php to invalidate all prior cache entries (see
> DefaultSettings.php).

I had the same problem as the OP, and setting wgCacheEpoch did not
seem to work for me (or maybe its effect was only delayed, due to some
timezone issue...). As my wiki is quite small, I wrote a script to
request all wiki pages with ?action=purge.

In a nutshell, the following commands should do the trick:

  host=http://wiki.my.org
  wiki=/wiki/

  wget -qO- "${host}${wiki}Special:Allpages" |
  sed -re '/start content/,/end content/p' -e 's/href="([^"]+)"/\nGREPME:\1\n/g' |
  sed -nre 's/^GREPME:(.*)$/\1/p' |
  grep "^${wiki}" |
  while read page
  do 
    wget -O/dev/null "${host}${page}?action=purge"
  done

Here is the complete script, with some boilerplate command line
processing.

#! /bin/bash
# mwpurgecache - invalidate mediawiki cache

set -e

PROG=$(basename $0)
VER=0.1

function version() {
  echo "$PROG $VER - invalidate mediawiki cache
Copyright (c) 2005 Claudio Jolowicz <claudio at jolowicz.com>

Copying and distribution of this file, with or without modification,
are permitted in any medium without royalty provided the copyright
notice and this notice are preserved. This programs comes with
ABSOLUTELY NO WARRANTEE.
"
}

function usage() {
  version
  echo "Syntax: $PROG [options] [--] [pages]

Options:

  -H, --host          host name           (default: localhost)
  -w, --wiki WIKI     server path to wiki (default: /wiki/)
  -u, --user USER     username for HTTP authentication
  -p, --passwd PASS   password for HTTP authentication
  -d, --dryrun        show commands without executing them
  -q, --quiet         quiet (no output)
  -v, --verbose       be verbose (print server response)
  -h, --help          display this message
  -V, --version       display version information

Options and arguments must be separated by whitespace.  WIKI must
match wgArticlePath, such that \$wgArticlePath = \"WIKI/\$1\". WIKI
must not contain regex special characters.

This script requests all wiki pages listed on the command line by
HTTP, prefixed by the wiki path, appending \`?action=purge' to the
URL. If no pages are specified, it retrieves the list of all pages
from Special:Allpages.

Examples:

  \$ $PROG Main_page Help:Contents
  \$ $PROG -H wiki.my.org -w /mediawiki -u joe -p banana
  \$ $PROG -H wiki.my.org --dryrun | wc -l
"
}

## The '-e#' is a noop option to avoid `Unsupported scheme' errors.
user='-e#'
passwd='-e#'
run=
verbose=-nv
quiet='-e#'

opt_host=http://localhost
opt_wiki=/wiki/
opt_user=
opt_passwd=
opt_dryrun=false
opt_quiet=false
opt_verbose=false
while [ "$1" ]
do
  case "$1" in
    -H|--host) opt_host="$2"; shift;;
    -w|--wiki) opt_wiki="$2"; shift;;
    -u|--user) opt_user="$2"; shift;;
    -p|--passwd) opt_passwd="$2"; shift;;
    -d|--dryrun) opt_dryrun=true;;
    -q|--quiet) opt_quiet=true;;
    -v|--verbose) opt_verbose=true;;
    -h|--help) usage; exit 0;;
    -V|--version) version; exit 0;;
    --) shift; break;;
    -*) usage 2>&1; exit 1;;
    *) break;;
  esac
  shift
done
opt_host=${opt_host%/}
opt_wiki=/${opt_wiki#/}
opt_wiki=${opt_wiki%/}/
[ x"$opt_user" = x"" ] || user="--http-user=${opt_user}"
[ x"$opt_passwd" = x"" ] || passwd="--http-passwd=${opt_passwd}"
$opt_dryrun && run=echo
$opt_verbose && verbose="--server-response"
$opt_quiet && quiet="-q"
(
  if [ $# -eq 0 ]; then
    wget -qO- "$user" "$passwd" "${opt_host}${opt_wiki}Special:Allpages" |
    sed -re '/start content/,/end content/p' -e 's/href="([^"]+)"/\nGREPME:\1\n/g' |
    sed -nre 's/^GREPME:(.*)$/\1/p' |
    grep "^${opt_wiki}" 
  else
    for page; do echo "${opt_wiki}${page}"; done
  fi
) | 
while read page
do
  $run wget -O/dev/null "$user" "$passwd" "$quiet" "$verbose" "${opt_host}${page}?action=purge"
done

exit 0

-- 
Claudio Jolowicz <claudio at jolowicz.com> | http://www.jolowicz.com
ICQ# 225314573 | Jabber: joloc at amessage.de | IRC: (freenode)joloc
1024D/C61DFB0C AA9A F078 2FD5 3459 D071  4998 D607 4FEA C61D FB0C
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.wikimedia.org/pipermail/mediawiki-l/attachments/20050818/bb795f3b/attachment.pgp 