Semi Protection

UESPWiki:Upgrade History/2012

The UESPWiki – Your source for The Elder Scrolls since 1995
Jump to: navigation, search
This is an archive of past UESPWiki:Upgrade History discussions. Do not edit the contents of this page, except for maintenance such as updating links.

29 December 2012

  • Removed Icon extension on content 1/2/3 (functionality now built in to MediaWiki software). Robin Hood  (talk)
  • Ran updateCollation.php to fix category "Next 200" issues. Robin Hood  (talk)
  • Updated SiteCustomCode.php on content1/2/3 to extend Blockers' blocking time from 4 hours to 6. Robin Hood  (talk)

17 December 2012

  • Updated MetaTemplateCategoryPage.php, SiteSpecialWantedpages.php, and SiteCustomCode.php form content3 to content1/2 to fix several bugs from the upgrade.

16 December 2012

  • Updated the mobile site ( to the current MediaWiki version as well as the WPTouch skin. Changed settings for the mobile sites to go through Squid and be server via the content servers as normal. Mobile wiki location is in /home/uesp/www/mobile/w/.
  • Cleared free space on content3 by deleting old logs.
  • Added mobile link in sidebar. Directs to which takes the referring URL and redirects to either the regular or mobile site depending on the referrer's URL (converts from the regular site to the mobile one).

14 December 2012

  • Explicity prevented Random from being cached by Squid on squid1.

9 December 2012

  • Fixed missing srmap handler in the UespMap extensions which permits the Skyrim map to be edited.
  • Manually copied the SRData_* databases from db1 to db2 to prevent replication from breaking when they are modified on db1.
  • Copied CSData_* permissions from content3 to db1 and db2.
  • Added CSData_* databases to backups on db1/db2/content3.

6 December 2012

  • Re-ran updateRestictions.php maintenance script to clear doubly-protected pages in MediaWiki space.

2 December 2012

  • Deleted old logs/files on db1 to free up space.
  • Added the "Dragonborn" namespace to content1/2/3 wikis.

26 November 2012

  • Added site notice for the end of the survey period.
  • Cleared Squid cache of most content pages to force update of site notice.

15 November 2012

  • Disabled daily backups on db1 as they were redundant and causing a 30 minute site outage each morning when they are run. A full weekly backup on db1 is still being done.

16 October 2012

  • Rebooted content2 to see if it solves the random connection issues which persists even though it was not being used for the past week.

12 September 2012

  • Added the SimpleAntiSpam extension to all content servers.

11 September 2012

  • Added editinterface permission to the patrollers groups.
  • Made the default timezone for anonymous users GMT.
  • Added a daily script on all content servers to clear /tmp of image related files once a day (files older than one day are deleted).
  • Manually updated the $wgCacheEpoch on all content servers to 10 Sept 2012.

7 September 2012

  • Blocked WarTune and R2Games from advertising due to a complaint and questionable content.

6 September 2012

  • Updated to MediaWiki v1.19.

5 September 2012

  • Turned off wgMainCacheType, wgParserCacheType, and wgMessageCacheType on content1/2 due to an issue with these interfering with gadgets.

4 September 2012

  • Restored ReCaptcha operation on content2/3.
  • Created missing external.png image.
  • Changed a reference to Namespace:: to the correct MWNamespace:: in the MetaTemplate extension on content1/2/3.
  • Changed to a manual wiki cache invalidation on content1/2/3 by setting $wgInvalidateCacheOnLocalSettingsChange = false and $wgCacheEpoch to a set value. This prevents a load spike whenever the LocalSettings.php file is changed. For major changes the $wgCacheEpoch should be slowly increased to minimize the lag spike due to cache invalidation.
  • Made change to /w/includes/logging/PatrolLog.log according to this bug report to fix the issue of patrol edits being attributed to the wrong person.
  • Set $wgUseXVO = true on content1/2/3.
  • Turned off the wiki file cache on content1/2/3 as it was preventing the Squid cache from caching any page since the upgrade. When the file cache is on the cache control header revert to private max-age=0 instead of the usual s-maxage=..., max-age=.....
  • Used $wgResourceLoaderMaxage to change the maxage of all unversioned resources from the default 5 minutes to 1 day.

3 September 2012

  • Upgraded wiki to v1.19 on content1/2/3.
  • Deleted the wiki file cache.
  • Restarted memcached on files1. Note that the memcached stats before restarting were:
  STAT pid 8200
  STAT uptime 9850289
  STAT time 1346684250
  STAT version 1.4.5
  STAT pointer_size 32
  STAT rusage_user 4147.842432
  STAT rusage_system 9695.774018
  STAT curr_connections 10
  STAT total_connections 54294602
  STAT connection_structures 431
  STAT cmd_get 186204755
  STAT cmd_set 13003461
  STAT cmd_flush 0
  STAT get_hits 175240610
  STAT get_misses 10964145
  STAT delete_misses 24623
  STAT delete_hits 268620
  STAT incr_misses 54
  STAT incr_hits 92491723
  STAT decr_misses 0
  STAT decr_hits 0
  STAT cas_misses 0
  STAT cas_hits 0
  STAT cas_badval 0
  STAT auth_cmds 0
  STAT auth_errors 0
  STAT bytes_read 23630315967
  STAT bytes_written 2049004619510
  STAT limit_maxbytes 805306368
  STAT accepting_conns 1
  STAT listen_disabled_num 0
  STAT threads 4
  STAT conn_yields 0
  STAT bytes 545986439
  STAT curr_items 422475
  STAT total_items 13002451
  STAT evictions 4057
  STAT reclaimed 3388561
  • Restored ReCaptcha operation on content1.

2 September 2012

  • Changed the wiki session name on content1/2 back to the original "uesp_net_wiki5".

30 August 2012

  • Changed $wgAutoblockExpiry from the default of 86400 to 7776000 seconds (3 months) on content1/2/3.

4 August 2012

  • Added missing Google analytics script/tag to wiki pages. Cleared wiki and squid caches.

2 August 2012

  • Changed all ads to use Google's DoubleClick For Publishers service (DFP). The only thing that should change is slightly improved ad loading performance (doesn't affect the page load time as much).
  • Clear the Wiki file cache and most of the Squid cache to force the ad changes to propagate faster.
  • Moved the Google+ link below the search text box.

30 July 2012

  • Added reverse DNS entries to some sub-domains to prevent outgoing mail from bouncing at some mail hosts (for example, forum/wiki activation e-mails). Confirmed working by relevant error messages in mail log no longer occurring regularly.

28 July 2012

  • Cleared most of the Squid cache to get rid of old/invalid cached pages.
  • Added Google Plus icon to Wiki pages (location may change).

26 July 2012

  • Prevented the temporary MediaWiki 1.19 directories from being publicly available (bots or random users may have been causing cache corruption by accessing them).
  • Deleted the Wiki file cache. Will still need to clean out the Squid cache for all pages that use a v1.19 MediaWiki instead of the current v1.14.

21 June 2012

  • Deleted a bunch of old/unused DNS entries that may have been causing a few random issues here and there.

19 June 2012

  • Fixed backups to dump correct wiki database.
  • Removed occasional popup ad from "Scorecard Research".
  • Restored content3 to use correct v1.14 wiki database.

11 May 2012

  • Changed iterative backup on db to start at a uesp_net_wiki5.text.old_id of 927000 for faster iterative backups.
  • Created Wiki clone database uesp_net_wiki14 from last full and iterative backups:
  • Full clone from dump: 47 minutes
  • Load Iterative No-Text: 6 minutes
  • Delete Text(>900000): 10 sec
  • Load Text(>900000): 24 minutes
  • Backup Text(>927000): 2 sec
  • Delete Text(>927000): 1 sec
  • Load Text(>927000): minutes
  • Created standard db permissions for uesp_net_wiki14

10 May 2012

  • Added the "Online" namespace to content1/2/3.
  • Changed iterative backup on db to start at a uesp_net_wiki5.text.old_id of 900000.

7 May 2012

  • Deleted old MySQL bin-logs on content3 to free up disk space.

6 May 2012

3 May 2012

  • Changed the mobile wiki to use db1.
  • Turned off file caching on the mobile wiki.

2 May 2012

  • Upgraded PHP on content2/3 to v5.2.
  • Re-installed older version of wikidiff2 on content2.
  • Updated wikidiff2 on content3.

22 April 2012

  • Ran the ``updateRestictions.php`` maintenance script.

14 April 2012

  • Added allspacepatrol perms to autopatrolled group

6 April 2012

  • Added autopatrolled group
  • Cleaned up some errors in existing groups and perms

3 April 2012

  • Fixed polls not working on the forums.
  • Fixed forum link to ACP not working for admins.

26 March 2012

  • Fixed file permissions on content1.
  • Changed forum chat channel to #uespforums on
  • Fixed forum skin settings and tweaked/fixed appearance.
  • Upgraded to modern google custom search on the forums.

25 March 2012

  • Upgraded forums to phpBB 3.0.10
  • Added "Chatroom" link to forums using Mibbit

21 March 2012

  • Reset database replication on db2 again and removed unnecessary SUPER privileges from several MySQL users (SUPER bypasses the read only setting).

20 March 2012

  • Changed Wiki settings (wgRC2UDP...) on content1/2/3.
  • Installed and setup epic IRC client on content3.
  • Reconfigured firewall to permit default IRC ports (194, 6660-6669).
  • Reset database replication on db2 and set read_only to ON to prevent any direct queries from breaking the slave.

5 March 2012

  • Restarted memcached on files1.
  • Cleared wiki file cache on files1.

28 February 2012

  • Changed main DNS to point to squid1.

27 February 2012

  • Deleted and recreated the Squid cache directory on squid1.
  • Restarted MySQL on db1 to stop the query log file from being used which was filling up the disk (150GB and counting).

25 February 2012

  • MetaTemplateSaveData.php modified to change how UNIQ/QINU marker tags are handled (now being stripped completely from #save'd variables)
  • Fixed read permission for uesp's web root directory on content2

24 February 2012

  • getmaplocs.php on content1/content2 changed to read from db1 instead of db2 because updates made to db1 are not propagating to db2
  • changes made to srmap.js to allow door icons to be displayed (temporarily being served from instead of because of permission issues)
  • Updated wiki/file permissions on content1/2/3 and files1.
  • Re-updated permissions on select files to *not* break the wiki and forums.

23 February 2012

  • Updated database grants/password for mapsedit user
  • Restored original wiki $wgDBname setting on content1/2/3
  • Set $wgSessionsInMemcached to false -- makes it impossible for apps to share info through session data, and based on history from last November the change didn't accomplish anything in the first place
  • Replace else if (strpos($_SERVER['PHP_SELF']."srmap")) with else if (strpos($_SERVER['PHP_SELF'],"srmap")) (replace period with comma) in setmaplocs.php on content1/2 (not content3, since it doesn't even have new code).
  • Restore ob_map_data from last night's backup on db1 -- rows had been overwritten with Skyrim data.

21 February 2012

  • Uploaded revision to UESPCustomCode, in which addImageClear function allows "class=" as well as "clear=" options to be added to image tags. Uploaded to content1/2/3

16 February 2012

  • Manually dumped and restored the uesp_net_wiki5.mt_save_set table from db1 to db2 to hopefully solve the reoccurring slave errors on db2 to due data mismatch.

15 February 2012

  • Switched content3 to use db2 for reads.
  • Switched content2 to use db2 for reads.
  • Switched content1 to use db2 for reads.

14 February 2012

  • Restored running MySQL slave on db2 by a full backup from content3.

4 February 2012

  • Switch back to db1 for all wiki reads due to a number of page errors.

3 February 2012

  • Switch to using db2 for all reads.

31 January 2012

  • db2 Planned Maintenance Downtime
  • Optimize MySQL on db1 by running the tuner script (reduce per client memory some)
  • Enabled log-slave-updates on db1 to permit it to be used as a slave and master
  • Switch all Wiki reads to db2
  • Perform full backup on db1
  • Stop slave on db1 and content3 and reset content3 to use db1 as a slave (instead of db2)
  • Switch all Wiki reads back to db1
  • Switch all maps to use db1
  • Switch to read-only on wiki and forums
  • Flush logs on db2
  • Stop and reset slave on db1 (record position before resetting)
  • NOTE: For future slave/master switches also record the local relay log file and position (this makes starting replication in the future potentially easier)
  • Switch wiki to use db1 for all writes/reads
  • Switch forums to use db1
  • Test writes on content3
  • Turn off read-only on wiki and forums
  • Switch misc sites (blog, eqwiki, davewiki) to use db1
  • Turn off MySQL on db2 and see what breaks...(nothing yet)

9 January 2012

  • Tweaked UespCustomCode on content1/2/3 to add a <link rel="canonical"> tag to all wiki page headers

Prev: 2011 Up: Upgrade History Next: 2013