Search
j0ke.net Open Build Service
>
Projects
>
Apache
:
Modules
>
apache2-mod_mirrorbrain
> apache2-mod_mirrorbrain.changes
Sign Up
|
Log In
Username
Password
Cancel
Overview
Repositories
Revisions
Requests
Users
Advanced
Attributes
Meta
File apache2-mod_mirrorbrain.changes of Package apache2-mod_mirrorbrain (Revision 7)
Currently displaying revision
7
,
show latest
------------------------------------------------------------------- Sun Aug 23 19:21:00 CEST 2009 - cs@linux-administrator.com - MirrorBrain 2.9.2 - see http://mirrorbrain.org/docs/changes/#release-2-9-2-aug-21-2009 ------------------------------------------------------------------- Thu Jul 30 05:20:00 CEST 2009 - cs@linux-administrator.com - MirrorBrain 2.9.1 - see http://mirrorbrain.org/docs/changes/ for details ------------------------------------------------------------------- Fri Jun 5 14:09:40 CEST 2009 - cs@linux-administrator.com - MirrorBrain 2.8.1 - Python 2.6 compatibility fixes: - mb file ls --md5: use hashlib module, if available (this fixes a DepracationWarning given by Python 2.6 when importing the md5 module) - mb list: the --as option had to be renamed to --asn, because "as" is a reserved keyword in Python, and Python 2.6 is more strict about noticing this also in cases where just used as an attribute. - remove the b64_md5 function, which is no longer used since a while - mb file ls: - make the --md5 imply the --probe option - mb export: - when exporting metadata for import into a VCS, handle additions and deletions - update docs to point to new RPM packages in the openSUSE build service (in a repository named Apache:MirrorBrain). The formerly monolithic package has been split up into subpackages. - perl-Config-IniFiles added to the list of perl packages required by the scanner ------------------------------------------------------------------- Sat Apr 4 16:52:53 CEST 2009 - poeml@suse.de - rename package to mirrorbrain and split off some subpackages, so that dependencies are a little more precisely located where they're needed: - mirrorbrain-tools - mirrorbrain-scanner - apache2-mod_mirrorbrain ------------------------------------------------------------------- Tue Mar 31 02:21:41 CEST 2009 - poeml@suse.de - update to r6964: * mb export: implement a new output format, named "vcs". Can be used to commit changes to a subversion repository and get change notifications from it. See "mb help export" and0 http://mirrorbrain.org/archive/mirrorbrain-commits/0152.html ------------------------------------------------------------------- Mon Mar 30 02:27:40 CEST 2009 - poeml@suse.de - package renamed from apache2-mod_zrkadlo (the old name) to apache2-mod_mirrorbrain. ------------------------------------------------------------------- Mon Mar 30 02:17:12 CEST 2009 - poeml@suse.de - update to MirrorBrain 2.8 (r6949): * scanner: - Larger improvements, mainly with regard to the definition of patterns for files (and directories) that are to be included from scanning. Old, hardcoded stuff from the scanner has been removed. Now, excludes can be defined in /etc/mirrorbrain.conf by scan_exclude and scan_exclude_rsync directives. The former takes regular expressions and is effective for FTP and HTTP scans, while the latter takes rsync patterns, which are passed directly to the remote rsync daemon. See http://mirrorbrain.org/archive/mirrorbrain-commits/0140.html for details. This can decrease the size of the database (>20% for openSUSE), and for many mirrors it considerably shortens the scan time. - Fixed a bug where the scanner aborted when encountering filenames in (valid or invalid) UTF-8 encoding. See https://bugzilla.novell.com/show_bug.cgi?id=490009 - Improved the implementation of exclusions as well as the top-level-inclusion pattern, which were not correctly implemented to work in subdir scans. - The source file was incorporated into the tools subdirectory. * mb dirs: - New subcommand for showing directories that the database contains, useful to tune scan exclude patterns. * The documentation was enhanced in some places, including the installation docs.. * mod_autoindex_mb: - This module (which is based on mod_autoindex) was rebased on httpd-2.2.11. ------------------------------------------------------------------- Fri Mar 27 15:53:35 CET 2009 - poeml@suse.de - require Apache >= 2.2.6 and libapr-util1 >= 1.3.0 ------------------------------------------------------------------- Mon Mar 9 22:33:59 CET 2009 - poeml@suse.de - update to r6772: * mod_mirrorbrain: - documentation updates (mod_mirrorbrain.conf, the example) ------------------------------------------------------------------- Mon Mar 9 10:17:33 CET 2009 - poeml@suse.de - update to r6748: * scanner: - implemented deletion of files from the mirror database for subdirectory scans. Before, this was only possible with a full scan. The new database scheme made this possible. - fixed the implementation of the whitelist for top-level directories in combination with subdir scans. - increased the commit frequency for rsync scans, to hold transactions open less long. - fixed the "enable after scan" option for parallel scans * mb list: - add new options to display further selected info about mirrors: --country --region --prefix --as --prio The --country conflicts a little with the previous -c|--country option to select mirrors; the latter is -c now, which is a little dumb. ------------------------------------------------------------------- Wed Mar 4 18:39:08 CET 2009 - poeml@suse.de - update to r6694: * mb file: - files looked up in the database can now be probed for, and the response displayed, optionally md5 hash of the downloaded content. This can be used to check functionality of the mirrors. * mb new: - when a new mirror is created, automatically fill in AS number and prefix, as well as a little comment with timestamp * NEWS updated, and COPYING file added ------------------------------------------------------------------- Tue Mar 3 14:57:58 CET 2009 - poeml@suse.de - update to r6671: * mod_mirrorbrain: - make insertion of broken mirrors into metalinks more practical, by making it configurable with MirrorBrainMetalinkBrokenTestMirrors a directive which specifies a file containing an XML fragment to be inserted. - document the recently added "as=..." query parameter, which causes a request to be treated as if it was coming from the autonomous system with the given number * metalink-hasher: - make --ignore-mask work for directories, too * mb tool: - fix a namespace problem (?) with module importing - small fix for "mb delete" ------------------------------------------------------------------- Mon Mar 2 23:02:17 CET 2009 - poeml@suse.de - update to r6661: * mod_mirrorbrain: - add broken URLs to the top of metalinks for testing purposes. Clients can trigger this with additional HTTP request header "X-Broken-Mirrors: true" - remove the variables (in Apache's subprocess env) with overly long names that have been replaced with shorter versions a while ago ------------------------------------------------------------------- Mon Mar 2 12:46:54 CET 2009 - poeml@suse.de - update to r6649: - add directions for upgrading from MB 2.6 to 2.7 ------------------------------------------------------------------- Mon Mar 2 02:15:56 CET 2009 - poeml@suse.de - update to r6648: * mod_mirrorbrain 2.7: - New database scheme for PostgreSQL. Instead of a potentially huge relational table including timestamps (48 bytes per row), files and associations are now in a single table, using smallint arrays for the mirror ids. This makes the table 5x faster and 1/3 the size. In addition, we need only a single index on the path, which is a small and very fast b-tree. This also gives us a good search, and the chance to do partial deletions (e.g. for a subtree). * mb tool: - adjust "mb file [ls|add|rm]" and "mb delete" to the array-based database scheme - adjust "mb vacuum" as well. It's much easier and also very fast now. - mb mirrorlist: fix the mb.files.check_for_marker_files() function to work with the new mirr_hasfile_byname function * scanner (v0.30): - more work on the new database scheme; now fully implemented - group database writes and commit them in chunks of useful size - fix criminal indentation - pass down -I to spawned scanners - added a few more hardcoded directories to ignore in the openSUSE tree. Of course, this needs to be made cleanly configurable. - handle concurrent insertion of a new file by another scanner. When scanning for the first time (starting with an empty database) there is a certain likelyhood that this may occur. ------------------------------------------------------------------------ r6616 | poeml | 2009-03-01 21:42:43 +0100 (Sun, 01 Mar 2009) | 4 lines metalink-hasher: - add --ignoremask option that specifies a regexp for files that the hashes should not be done for0 ------------------------------------------------------------------- Thu Feb 26 15:45:28 CET 2009 - poeml@suse.de - update to r6597: * scanner: - improve FTP reconnection handling. Attempt to reconnect only when more than 60 seconds have passed since the last activity in the connection, thus dealing with a true timeout that happened in the FTP server. ------------------------------------------------------------------- Thu Feb 26 03:20:07 CET 2009 - poeml@suse.de - update to r6594: * scanner: - implement for FTP, what was done for HTTP and rsync yesterday: allow the definition of top-level directories that are scanned and ignore all others. Saves on database size, and makes scans a lot faster in many cases. - in addition, the configuration to ignore certain paths has never worked for FTP; should be fixed now as well, however I didn't test it. ------------------------------------------------------------------- Wed Feb 25 16:18:42 CET 2009 - poeml@suse.de - update to r6588: * scanner: - fix a two-fold bug in FTP scanning: First, ignore bogus return values from Perls Net::FTP, which claim a 550 status code when in fact a timeout happend (421). FTP servers may close connections when they are idle, and when the scanner (or database) takes to long to chew on a filelist, this may happen. Fix both issues by ignoring the reported 550, and attempting to reconnect. ------------------------------------------------------------------- Wed Feb 25 03:42:50 CET 2009 - poeml@suse.de - update to r6572: * scanner: - another measure to save on database size: allow the definition of top-level directories that are scanned, and all others ignored. Implemented for HTTP and rsync scanning so far; FTP missing. - add stub for using a different database scheme (experimental, disabled) * mb scan: - pick up a new option named scan_top_include from /etc/mirrorbrain.conf and pass it to the scanner * mb vacuum: - fix mb vacuum, the database cleanup tool, to work with PostgreSQL MySQL idiosyncrasies removed, (although I didn't test it with MySQL to see whether it still works with that database) ------------------------------------------------------------------- Tue Feb 24 20:18:08 CET 2009 - poeml@suse.de - update to r6564: * mod_mirrorbrain: - for logging purposes, distinguish mirrors that are in the same country from mirrors that act as fallback mirrors for the country * mb mirrorlist: - take care that the new private mirror type doesn't end up on the generated public mirror list ------------------------------------------------------------------- Tue Feb 24 17:38:53 CET 2009 - poeml@suse.de - update to r6561: * mod_mirrorbrain: - fix bug that occured with version 2.5 and affects PostgreSQL installs: make the region_only, country_only, as_only, prefix_only flags work again - allow to override the client IP's autonomous system by HTTP query parameter (?as=[0-9]*) - implement the missing redirection exceptions mirrors marked AS-only or prefix-only. - fix selection of mirrors into the metalinks, which didn't respect the *_only flags so far. - avoid rendering invalid XML for metalinks, when autonomous system or network prefix are unknown, and set to "--". Show "unknown" instead. ------------------------------------------------------------------- Sun Feb 22 16:31:45 CET 2009 - poeml@suse.de - update to r6537: * mb tool: - handle DNS errors when updating mirrors with network data * scanner: - better output, showing more exactly where time is spent. - when scanning all mirrors, start with those that have been scanned longest ago - cleanup of dead code ------------------------------------------------------------------- Sat Feb 21 20:35:59 CET 2009 - poeml@suse.de - update to r6531: - MirrorBrain SQL schema: * remove timestamp_file column * timestamp_scanner is a UNIX epoch now, which is a little smaller than an SQL timestamp. * better indexes - mod_mirrorbrain: * shorten the SQL query - scanner: * adjusted for the new SQL scheme * fix printing of database error strings - mb tool: * work around the requirement of SQLObject to have a primary key in each table, by simulating one for it (with a table view) * mb iplookup: add --all-prefixes option ------------------------------------------------------------------- Fri Feb 13 08:19:25 CET 2009 - poeml@suse.de - update to r6452: - mod_mirrorbrain 2.6 * additional mirror selection criteria are autonomous system and network prefix, as looked up by mod_asn and provided by the mirror database * shorten the names of variables set in the subprocess environment - the long ones are still there but can be obsoleted soon, once the change is deployed. * set MB_REALM to be available for logging. - Two new mirrordoctor tools added: * "mb iplookup" looks up network prefix and AS * "mb update" can set/update this data across the database (all mirrors) ------------------------------------------------------------------- Thu Feb 12 18:49:04 CET 2009 - poeml@suse.de - update to r6444: - mb tool: * allow to set the mirrorbrain instance via MB env var (command option still exists, and overrides it) ------------------------------------------------------------------- Thu Feb 12 02:09:43 CET 2009 - poeml@suse.de - update to r6441: - metalink-hasher: * make sure that the file and directory permissions are not copied, because for staged content this means that they might be unreadable. Since the hashes don't contain any reversible information, there is no reason to treat them as confidential (and Apache would not deliver them anyway, since they are not in the Docroot at all). Add an option that allows permission copying." tools/metalink-hasher.py ------------------------------------------------------------------- Thu Feb 12 01:02:55 CET 2009 - poeml@suse.de - update to r6440: - scanner: * fix missing initialization of a variable used for logging the server identifier - useful for debugging ------------------------------------------------------------------- Wed Feb 11 23:35:13 CET 2009 - poeml@suse.de - update to r6439: - Interim release with further work on PostgreSQL support - - updates to INSTALL docs - MirrorBrain SQL schema: * add country and region data for importing into newly created databases * add asn and prefix field, and flags to limit mirror selection by these, and remove obsolete netblock field - mb probefile: * add --md5 option; when doing probe requests, it'll do actual downloads and show md5 sums of the responses - mb export: * add postgresql (insert statements) as export format ------------------------------------------------------------------- Wed Feb 4 21:25:51 CET 2009 - poeml@suse.de - update to r6377: - scanner: * be less verbose, don't print out each directory visited. Rather show progress in a way that makes sense and stays readable when mirrors are scanned in parallel. Prefix output with mirror names ------------------------------------------------------------------- Wed Feb 4 17:30:50 CET 2009 - poeml@suse.de - update to r6373: - mod_mirrorbrain: * for reasons of consistency, remove the MirrorBrainInstance directive when the module is compiled without memcache support, because it was used only to give memcache keys a unique name. - mirrorprobe: * be less verbose: don't send mail, and don't add a note to the comment field each time a mirror is not reachable * make it work with PostgreSQL ------------------------------------------------------------------- Tue Feb 3 23:22:36 CET 2009 - poeml@suse.de - update to r6360: - scanner: - fix enabling servers after scan. Make it work with both PostgreSQL boolean type as well asn MySQL's tinyint. - mirrorbrain tool (mb delete): - when deleting a mirror, cascade delete all references first - further work on the INSTALL document ------------------------------------------------------------------- Tue Feb 3 21:10:10 CET 2009 - poeml@suse.de - update to r6357: - scanner: 0.22 - more efficient SQL statement handling - added SQL logging option for debugging - mb (mirrorbrain tool): - bugfix in the file command: make patterns work which have a wildcard as first character. - extend "mb scan" to accept -v and --sql-debug and pass it to the scanner ------------------------------------------------------------------- Tue Feb 3 10:49:06 CET 2009 - poeml@suse.de - update to r6332: version 2.5 * intermediate release - working on the database schema, and on the tools, to allow PostgreSQL as backend database. * working on the installation howto ------------------------------------------------------------------- Fri Jan 23 23:05:39 CET 2009 - poeml@suse.de - update to r6154: version 2.4 * mod_zrkadlo was renamed to mod_mirrorbrain * memcache support is now optional at compile time * update installation instructions ------------------------------------------------------------------- Fri Jan 23 15:39:17 CET 2009 - poeml@suse.de - update to r6128: * mod_zrkadlo: - accept digits in two-letter country codes, so that satellite "country" A2 is accepted in queries ------------------------------------------------------------------- Wed Jan 21 20:33:09 CET 2009 - poeml@suse.de - update to r6075: * mod_zrkadlo version 2.3: - rely on mod_geoip to do GeoIP lookups. Deployment obviously equires configured mod_geoip. - deprecate "clientip" query parameter, which can no longer work once we use mod_geoip. - implement 'country" parameter that can be used instead. ------------------------------------------------------------------- Wed Jan 21 01:17:10 CET 2009 - poeml@suse.de - update to r6065: * mod_zrkadlo: - don't do memory-intensive caching of the GeoIP database. * geoiplookup_continent: - make continent lookups work with GeoIPCity databases * geoip-lite-update tool: - adjust for the new URL of GeoLiteCountry database - download also updates of the city database * geoiplookup_city: - new tool to show details from the GeoIP city databases ------------------------------------------------------------------- Mon Dec 29 23:35:48 CET 2008 - poeml@suse.de - update to r5980: * mirrorprobe: - auto-reenable mirrors. - include details in the mail subject, not only in the body * mb markers: - new commandline tool to edit marker files in $EDITOR * mb probefile: - add option to probe on a single mirror * mb mirrorlist: - allow configurable html footer & header to be added in - override caption and title by commandline arguments. - remove openSUSE specific stuff that was hardcoded - add timestamp to header template - get regions from the database - improve table - odd row class, cgi-escape html entities, ... - every 20 rows, add another table header line - exclude wildcard mirrors from the mirror lists. - use stars instead of the raw priority values - allow to add the creation time into the templates - assign <col id="..."/> column ids - add title attributes into each td element for accessibility (not sure whether the column ids/headers are enough) * tools/create_timestamp: add an explanatory text to the timestamp files * change "Taiwan, Province of China" to the more neutral "Taiwan" ------------------------------------------------------------------- Sat Dec 13 14:54:16 CET 2008 - poeml@suse.de - update to r5892: * mb probefile: - add support for checking not only HTTP, but also FTP and rsync URLs. - bypass intermediate caching proxies. - use a rigorous timeout of 20 seconds for every network operation. - add option to also test mirrors which are currently disabled, or whose HTTP server is unreachable. * mirrorprobe: - handle refused connections on FTP URLs - use GET requests rather than HEAD requests. It is safer; I just saw a crashed mirror hanging indefinitely only on GET requests. * rsyncusers tool: - make it compatible with oldish Python (2.3) * create_timestamp: - a new tool added. Can serve as example. ------------------------------------------------------------------- Thu Dec 4 05:05:49 CET 2008 - poeml@suse.de - update to r5802: * mb tool: add editable public_notes field ------------------------------------------------------------------- Tue Dec 2 03:34:41 CET 2008 - poeml@suse.de - update to r5778: * mb tool: work on generated mirrorlists. ------------------------------------------------------------------- Tue Nov 25 13:46:35 CET 2008 - poeml@suse.de - update to r5701: * mod_zrkadlo: improve the wording in mirror lists, so it doesn't claim that fallback mirrors are in the same country ------------------------------------------------------------------- Sat Nov 22 10:59:43 CET 2008 - poeml@suse.de - update to r5679: * scanner: make the usage of md5 hashes optional, and disable it by default * mb tool (mirrordoctor): rework file subcommand * remove md5 support from mod_zrkadlo ------------------------------------------------------------------- Tue Nov 18 00:25:28 CET 2008 - poeml@suse.de - update to r5623: add interim support for queries that don't use the extra hashes to look up files in the database, but access them directly by filename. This could eventually serve to get rit of the extra hash column, and let a database index do the job. ------------------------------------------------------------------- Sun Nov 16 15:24:42 CET 2008 - poeml@suse.de - added shortcut 'mb' for the longish 'mirrordoctor' command ------------------------------------------------------------------- Sun Nov 9 19:24:09 CET 2008 - poeml@suse.de - update to r5573 * mod_zrkadlo: avoid false compile warning about comparison with string literal - compile geoiplookup_continent tool with RPM_OPT_FLAGS ------------------------------------------------------------------- Sun Nov 9 18:03:01 CET 2008 - poeml@suse.de - update to r5570: mod_zrkadlo release 2.1: * simplified the Apache configuration: It is no longer needed to configure a database query. At the same time it's less error-prone and avoids trouble if one forgets to update the query, when the database schema changes. * specific mirrors can be now configured to get only requests for files < n bytes ------------------------------------------------------------------- Mon Nov 3 17:47:08 CET 2008 - poeml@suse.de - update to r5506: smallish mirrordoctor updates: * when listing files matching a pattern, show actual filename, not the pattern * add support for editing of the new other_countries database field * add first stub for mirror tests ('test' subcommand) ------------------------------------------------------------------- Mon Nov 3 12:43:42 CET 2008 - poeml@suse.de - update to r5498: mod_zrkadlo release 2.0: * implement better fallback mirror selection. Fallback mirrors are now defined in the database (server.other_countries column) and used only when no local mirror could be found (fallback). The obsolete ZrkadloTreatCountryAs directive has been removed. * double-check for empty base URLs of mirrors ------------------------------------------------------------------- Wed Oct 29 16:24:03 CET 2008 - poeml@suse.de - update to r5469: * new subcommand 'mirrordoctor file' for file operations on the mirror database. It can list files add add/remove them. ------------------------------------------------------------------- Tue Oct 28 23:27:20 CET 2008 - poeml@suse.de - update to r5465: * new subcommand 'mirrordoctor vacuum' for database pruning. Incorporates the functionality of previous /usr/bin/vacuum. ------------------------------------------------------------------- Mon Oct 27 12:47:27 CET 2008 - poeml@suse.de - update to r5445: * add geoip-lite-update, a script to automate updates of the GeoIP database ------------------------------------------------------------------- Mon Oct 27 10:22:49 CET 2008 - poeml@suse.de - update to r5442: * fix link to bittorrent resources. ------------------------------------------------------------------- Mon Oct 27 03:46:23 CET 2008 - poeml@suse.de - update to r5441: mod_zrkadlo version 1.9: * Check for files ending in .torrent, and if one exists, add a link to it into generated metalinks. New directive ZrkadloMetalinkTorrentAddMask, which can used to limit this additional check to specific files or directories. * rewrote HTML mirror list * add configurable CSS stylesheet to mirror lists (ZrkadloMirrorlistStyleSheet directive) ------------------------------------------------------------------- Fri Oct 24 23:14:38 CEST 2008 - poeml@suse.de - update to r5427: * metalink-hasher: add PGP signatures into <verification> block ------------------------------------------------------------------- Wed Sep 24 01:33:13 CEST 2008 - poeml@suse.de - update to r5127: * mirrorprobe: - make sure that we don't try to access non-existing attribute named code when it doesn't exist (like for FTP urls) ------------------------------------------------------------------- Fri Sep 19 21:10:38 CEST 2008 - poeml@suse.de - update to r5090: * rsyncusers: - new, performant, implementation in Python. Uses Python generators, which means it allocates only few memory and works similar a Unix pipe. The tool will now transparently open uncompressed, gzip or bzip2 compressed files. ------------------------------------------------------------------- Tue Sep 2 15:48:37 CEST 2008 - poeml@suse.de - update to r4855: * mod_zrkadlo: - implement the redirection exceptions (file too small, mime type not allowed to be redirected etc) for transparently negotiated metalinks. - add Vary header on all transparently negotiated resources. - remove depracated negotiation on Accept-Features. ------------------------------------------------------------------- Fri Aug 29 22:36:23 CEST 2008 - poeml@suse.de - update to r4830: * scanner: - more fixes to deal with more variants of the HTML autoindex breed. - append ?F=1 to HTTP requests (autoindex modifier) * mirrorprobe: - add the name of the mirrorbrain instance to warning mails * mod_autoindex_mb: if metalink or mirrorlists are active, add our signature * update INSTALL for creating new mirrors ------------------------------------------------------------------- Thu Aug 28 15:44:24 CEST 2008 - poeml@suse.de - update to r4800: * metalink-hasher: - don't drop the <pieces> elements. Our metalinks were lacking the length of hash pieces due to this bug. Clients were probably ignoring the partial hashes. * mirrordoctor list: - add --disabled option to show only disable mirrors * mirrordoctor scan: - leave a comment after successful scanning and enabling - add -a|--all option to pass to the scanner - remove -f option which is redundant somehow * mirrordoctor new: - improve usage. The -i option is no longer needed, the identifier is simply given as argument to the command. This is more in line with the other subcommands. - don't print SQL when not in debug mode. ------------------------------------------------------------------- Wed Aug 27 12:47:13 CEST 2008 - poeml@suse.de - update to r4788: * mirrordoctor rename: fix the call which looks up mirrors. mirrordoctor new: set default values for 'enabled' and 'status_baseurl', for tables that don't do it * mirrorprobe: add newlines before appending stuff to comments; don't try to remove stuff from comments, since multiline comments are now commonly used. ------------------------------------------------------------------- Mon Aug 25 14:58:40 CEST 2008 - poeml@suse.de - update to r4765: * mirrorprobe: fix for another irregularity caused by Nginx web servers * scanner: fix forking with multiple instances * metalink-hasher: add missing quoting ------------------------------------------------------------------- Sat Aug 23 01:01:03 CEST 2008 - poeml@suse.de - update to r4758: scanner improvements: - enable to parse HTML table autoindexes. - be more robust when detecting autoindexes. mirrordoctor fixes: - add option to specify path to scanner program - fix calling the scanner when no instance has been specified - add python-cmdln to Requires ------------------------------------------------------------------- Fri Aug 22 19:24:59 CEST 2008 - poeml@suse.de - update to r4753: mirrorprobe: include the instance name into the log messages ------------------------------------------------------------------- Fri Aug 22 18:18:02 CEST 2008 - poeml@suse.de - update to r4752: - pingd was renamed to "mirrorprobe". And it also deals with multi mirrorbrain instances now. In addition, it ignores proxy settings now. - simplified interface of the mirrorbrain python module (mb) - updated INSTALL docs ------------------------------------------------------------------- Thu Aug 21 20:10:36 CEST 2008 - poeml@suse.de - update to r4731: * mod_zrkadlo: - add ZrkadloInstance directive that can be set per mirrorbrain instance. To be set in server context, thus each instance needs a virtual host. - include the instance name into memcache keys - add new reconnect=0 parameter (that comes with apr-util-1.3.3) to mysql driver example configuration (prepared statements are gone after reconnection so that it does not make sense to reanimate dead connections. See https://issues.apache.org/bugzilla/show_bug.cgi?id=45407) * scanner: - make scanner work with multiple mirrorbrain instances (-b option) - fixed bnc394470: added 302, 303 and 307 codes to the check * mirrordoctor: - make it work with multiple mirrorbrain instances - list: hide disabled mirrors by default - add a rename subcommand - when disabling a mirror, set statusBaseurl to zero, so when re-enabling the mirror it doesnt' matter if it is not actually online. * pingd: - make sure that the mirror objects always have the response and response_code attributes set * SQL schema changes: - ensure that the identifier column of the mirror table doesn't allow duplicates - set a default value for the 'enabled' and 'status_baseurl' columns in the mirrors table ------------------------------------------------------------------- Fri Jun 20 16:03:55 CEST 2008 - poeml@suse.de - update to r4228: * mirrordoctor update: - make the commandline simpler -- now it is possible to just give a substring of the mirror identifier, instead of the full name (or -m) - add more options to the 'scan' command - add 'score' command - add setup.py file to make the the module installable ------------------------------------------------------------------- Wed Jun 18 18:06:22 CEST 2008 - poeml@suse.de - update to r4210: * mirror probe (ping script): fix crash when encountering a server which replies with broken chunked transfer encoding. nginx/0.6.29 does that. See http://bugs.python.org/issue1205 ------------------------------------------------------------------- Mon Jun 16 00:48:25 CEST 2008 - poeml@suse.de - update to r4178: * added mirrordoctor, a new commandline tool to maintain mirror entries in the redirector database. * build with aprutil 1.3, which ships apr_memcache. Thus, the latter is not separately available (and required) anymore. - Change the code which fetches rows from the query result set to do this in random access mode. Although we actually just want to access them sequentially, we need to do it this way because (at least at with the mysql driver) apr_dbd_get_row() doesn't return anything useful anymore if apr_dbd_num_tuples() has been called before. With apr-util 1.2.x, it worked nevertheless, but it doesn't with 1.3. * For wildcard mirrors (Content Delivery Networks), set their country and region to that of the client, so it turns up as useful data which the client understands in mirrorlists or metalinks. - enhance the metalink-hasher tool in two ways: - make it possible to run it on a subdirectory only - make it possible to work with different base directories -- which is useful if content is living in a stage area, but you want to create the hash pieces already. - rsyncusers tool: fix invalid shell code (spaces around '=') - add example configuration snippet for mod_memcache ------------------------------------------------------------------- Tue Jun 10 10:33:51 CEST 2008 - poeml@suse.de - build service supports the debuginfo flag in metadata now; remove debug_package macro from the specfile therefore. ------------------------------------------------------------------- Mon Jun 9 14:57:13 CEST 2008 - poeml@suse.de - update to r4133: - rsyncinfo: add option -S to hide rsync's stderr output ------------------------------------------------------------------- Mon Jun 2 17:35:22 CEST 2008 - poeml@suse.de - update to r4074 (version 0.8): - use mod_memcache for the configuration and initialization of memcache - metalink-hasher script added, to prepare hashes for injection into metalink files - rsyncusers analysis tool added - scanner bugfixes regarding following of redirects ------------------------------------------------------------------- Wed May 21 10:56:57 CEST 2008 - poeml@suse.de - update to r4011: - rsyncinfo: fix parsing of rsync module list, when no banner is prefixed. ------------------------------------------------------------------- Tue May 20 14:54:19 CEST 2008 - poeml@suse.de - update to r4002: - rsyncinfo: tabulate the output ------------------------------------------------------------------- Thu May 15 00:58:43 CEST 2008 - poeml@suse.de - don't package build intermediaries from mod_autoindex ------------------------------------------------------------------- Fri May 2 18:49:52 CEST 2008 - poeml@suse.de - update to r3877: - implement a testbed for testing failover with clients using text mirror lists; the server will inject broken mirrors at the top of the list if the user-agent is 'getPrimaryFailover-agent/0.1' - add rsyncinfo, script to get info on modules of rsync servers ------------------------------------------------------------------- Wed Apr 30 19:33:06 CEST 2008 - poeml@suse.de - update to r3874: - It was agreed that the new Metalink spec (3.0) will probably use RFC 3339 date and time format, but continue to use plain old RFC822 format. - add ZrkadloMetalinkPublisher directive to make name and url in publisher data configurable ------------------------------------------------------------------- Wed Apr 30 14:49:28 CEST 2008 - poeml@suse.de - update to r3871: - fix issue with requesting metalinks by appending ".metalink": the <size> element in metalinks was not filled in correctly anymore. ------------------------------------------------------------------- Sun Apr 27 23:12:35 CEST 2008 - poeml@suse.de - update to r3844: - now there is another, more natural, way to request a metalink: by appending ".metalink" to the filename - mod_autoindex_mb now creates links with .metalink appended, instead of ?metalink ------------------------------------------------------------------- Sun Apr 27 12:10:39 CEST 2008 - poeml@suse.de - update to r3840: - also return metalinks to clients which send application/metalink+xml in the Accept request header. ------------------------------------------------------------------- Mon Apr 21 18:17:24 CEST 2008 - poeml@suse.de - update to r3766: - mod_zrkadlo 1.7 - implement a terse text mirrorlist (suitable for e.g. http://en.opensuse.org/Libzypp/Failover) - allow clients to use RFC2295 Accept-Features header to select variants (metalink or mirrorlist-txt) - implement ZrkadloMetalinkHashesPathPrefix directive, for out-of-tree placement of prepared metalink hashes - work on unifying configuration handling of the tools, and make them all use /etc/mirrorbrain.conf ------------------------------------------------------------------- Sat Apr 19 16:52:40 CEST 2008 - poeml@suse.de - update to r3752: - metalinks: - add RFC 2183 header with filename + ".metalink" - avoid putting -- (from failed GeoIP lookup) into XML comments, since they will be seen as comment terminators - add mod_autoindex_mb.c which adds metalinks and mirrorlists to generated indexes ------------------------------------------------------------------- Fri Apr 18 19:59:38 CEST 2008 - poeml@suse.de - update to r3749: - bypass the redirection exceptions for metalinks, just as for mirrorlists ------------------------------------------------------------------- Thu Apr 17 18:17:29 CEST 2008 - poeml@suse.de - update to r3738: - make the scanner cope with servers that don't support range requests - small documentation updates ------------------------------------------------------------------- Mon Apr 7 23:46:19 CEST 2008 - poeml@suse.de - update to r3664: - updated THANKS file - some documentation updates in ABOUT file - remove useless header from mod_zrkadlo response - don't include the server_admin email address into the metalink - update metalink generator name - actually install the new THANKS file ------------------------------------------------------------------- Fri Apr 4 18:08:35 CEST 2008 - poeml@suse.de - update to r3634: - version 1.6 - implement metalink support (http://www.metalinker.org) - move the block where the single mirror is selected for redirection to after the mirrorlist/metalink if branch. It is not needed to do that work when metalink or mirrorlist is generated. - fix a bug I just discovered, but which has been existing since ever. I didn't allocate memory for mirror_entry's country_code member. - seed the RNG with getpid() instead of time(), because if the server is started several processes are forked off and get the same seed. Even though it doesn't play a real role in production, it is confusing when working in a debug setting (with many fresh starts). - THANKS file added ------------------------------------------------------------------- Fri Mar 28 17:31:56 CET 2008 - poeml@suse.de - update to r3592: - updated install documentation - scanner bugfixes, and HTTP large file capability check - fixed http_readdir: passes correct urls to largefile_check - scan tests with -j <n>, single servers, -a with mem verification -- some aborts in single jpobs occured due to rsync errors -- no infinite hangs - bugfix: wrong/broken urls were passed to the largefile sanity check thus creating false positives - http redirects 301 to ftp urls are treated as valid ------------------------------------------------------------------- Thu Mar 6 19:23:00 CET 2008 - poeml@suse.de - update to r3441: - scanner: get rid of hardcoded database credentials - integration of sanity checks for large files: large files visible on rsync/ftp which can't be downloaded via http are not put to the database anymore. Thanks, Jan! - set_mirror_score: also use the central conf file ------------------------------------------------------------------- Thu Mar 6 18:25:54 CET 2008 - poeml@suse.de - add perl-Config-IniFiles to requires ------------------------------------------------------------------- Thu Feb 28 13:57:53 CET 2008 - poeml@suse.de - add package requires for mod_form, GeoIP, python-mysql, python-sqlobject ------------------------------------------------------------------- Thu Feb 28 08:20:36 CET 2008 - poeml@suse.de - update to r3189: - the last change (version bump) wasn't actually probagated, because I forgot to re-spin the tarball. Here it is. - install the tools ------------------------------------------------------------------- Wed Feb 27 19:33:24 CET 2008 - poeml@suse.de - update to r3188: - version 1.5 - updated documentation - mirrorlist fixes: - ensure that any "clientip" parameter from query string represents a valid IP address, in order to keep generated HTML clean, so not to allow XSS attacks - set character set for HTML output generated to ISO-8859-1 - fix mirrorlist for the case where no mirror is found. Now an empty list is created, instead of sending the file in question. - when user asks for mirror list, don't DECLINE if the file would normally being excluded from redirection by one of the various criteria (size, user agent, ...) - use GeoIP_id_by_addr() instead of the deprecated (acc. to GeoIP.h) GeoIP_country_id_by_addr() ------------------------------------------------------------------- Wed Feb 13 16:28:58 CET 2008 - poeml@suse.de - update to r3124: - add details about the request to the subprocess_env table, so then can be logged via a CustomLog: - ZRKADLO_FILESIZE, the size of the file which was requested - ZRKADLO_COUNTRY_CODE and ZRKADLO_CONTINENT_CODE - if a mirror's baseurl doesn't end in '/', append it. Otherwise a non-working redirect URL could result. ------------------------------------------------------------------- Thu Jan 24 20:40:43 CET 2008 - poeml@suse.de - update to r3056: - version 1.4 - refined mirror selection by observing two additional server attributes in the database: country_only and region_only. - country_only=1 means that a mirror is not willing to receive requests from countries other than its own. - region_only=1 means that a mirror is not taking redirects from other continents, likewise. This is good for countries with poor internet connectivity. ------------------------------------------------------------------- Tue Oct 2 15:51:31 CEST 2007 - poeml@suse.de - update to r2298: - version 1.3 - implement ZrkadloExcludeNetwork directive, which allows for excluding client IPs from being redirected (simple string prefix comparison) - implement ZrkadloExcludeIP directive, which matches on IP address - replace apr_strnatcmp/apr_strnatcasecmp with their simpler glibc counterparts ------------------------------------------------------------------- Mon Oct 1 23:55:17 CEST 2007 - poeml@suse.de - update to r2289: - implement wildcard country support, useful for content delivery networks which should act as a mirror local to all countries ------------------------------------------------------------------- Thu Sep 20 20:57:55 CEST 2007 - poeml@suse.de - update to r2244: - version 1.2 - shorten the SQL query by looking up the base64 encoded md5sum of the filename in file_server.path_md5, avoiding an additional join on file.name. Needs an adjusted SQL query string, as shown in the example mod_zrkadlo.conf. ------------------------------------------------------------------- Fri Sep 14 04:17:19 CEST 2007 - poeml@suse.de - update to r2165 - version 1.1 - set ZRKADLO_REDIRECTED and ZRKADLO_NOMIRROR in r->subprocess_env, which can be used by a CustomLog - the "no mirrors found for %s" error log message is downgraded to APLOG_INFO - mod_zrkadlo.conf: give example for logging redirects, and files without any mirror - updated docs ------------------------------------------------------------------- Mon Aug 20 20:29:42 CEST 2007 - poeml@suse.de - update to r2012 - add new fields in statistic of memcached 1.2.x ------------------------------------------------------------------- Tue Mar 13 13:00:03 CET 2007 - poeml@suse.de - update to r1432 - bump version (1.0) - license under Apache License 2.0 - canonicalize the filename (resolve symlinks in the path) before looking it up in the database - small fixes in the mirrorlist output - remove debug messages about database connection acquisition ------------------------------------------------------------------- Tue Mar 6 07:35:14 CET 2007 - poeml@suse.de - update to r1387 - implement mirrorlist query string command to obtain a HTML list of all available mirrors - add ZrkadloMemcached directive to enable/disable memcache usage - add ZrkadloExcludeFileMask - unify logging ------------------------------------------------------------------- Sun Feb 25 14:08:32 CET 2007 - poeml@suse.de - update to r1348 - make memcached connection pool size configurable - make memcached object lifetime configurable ------------------------------------------------------------------- Wed Feb 21 02:26:05 CET 2007 - poeml@suse.de - add geoiplookup_continent ------------------------------------------------------------------- Tue Feb 20 16:53:49 CET 2007 - poeml@suse.de - package created