Last modified on 4 February 2017, at 19:55

User talk:Daveh/Archive 2010

This is an archive of past User talk:Daveh discussions. Do not edit the contents of this page, except for maintenance such as updating links.

User Group Permission TweakEdit

Merry Christmas Daveh. When you get a minute, please could you give the movefile right to the Bots user group? At the moment we have hundreds (probably thousands) of images with names that don't fit our naming conventions and it's a big job to fix this manually (and clogs up Recent Changes horribly). If I can get RoBoT to do the moves and tidy up afterwards it would be much easier. Thanks. –rpehTCE 15:04, 26 December 2009 (UTC)

Any news on this? –rpehTCE 21:01, 7 January 2010 (UTC)
Since you replied to another message, I'm guessing you're online ATM, so... bump. –rpeh TCE 01:31, 11 January 2010 (UTC)
I hate to bother you again, but please could you make this change next time you're around? Or, if you don't want to do it, tell me so I don't bother you about it any more. Thanks. rpeh •TCE 17:40, 9 February 2010 (UTC)
No worries...I just had forgot about the request and I don't mind being reminded. Done! -- Daveh 21:43, 13 February 2010 (UTC)
Thanks! Now I've just got to write the code to use it.... rpeh •TCE 12:07, 14 February 2010 (UTC)

Deleted Pages Not In Deletion LogEdit

I just did 3000-some-odd deletions, and I got pretty quick at it. Unfortunately, that may have caused a problem. We've got ten articles that are red-links, and have an edit history, but do not show in the deletion log. Here's some screenshots to illustrate:

The full list of pages with the issue follows:

Not sure what's going on, but something's definitely screwy. Also, the two that are showing as blue-links (Garlas Agea and Sanguine sub-pages) also give the same message as the third image, except not in the edit window (obviously). These are the two I attempted to undelete to address the issue; they were red-links before I did that. --GKTalk2me 21:46, 30 December 2009 (UTC)

I've just had a related issue and it seems User:Joram has too - let's continue over at UESPWiki:Administrator Noticeboard#Database Corruption. –rpehTCE 21:49, 30 December 2009 (UTC)
There are more, possibly-related problems. Update also at UESPWiki:Administrator Noticeboard#Database Corruption. rpeh •TCE 11:28, 28 January 2010 (UTC)



Could you please take a look at this problem?

While you're there I'd like you to apply this patch too. It removes an old workaround for Konqueror that causes modern webkit browsers to display the sidebar below the content area. I tried it with Konqueror 4, and the workaround isn't needed any more. -- Nx / talk 11:38, 8 January 2010 (UTC)

Sure, no problem. -- Daveh 02:47, 11 January 2010 (UTC)
Both done! -- Daveh 00:26, 12 January 2010 (UTC)
Thank you. -- Nx / talk 09:19, 12 January 2010 (UTC)

Removing the is_khtml variable breaks prefs.js and the tabbed preferences view no longer works. Just do this to fix it. My apologies, I didn't test the actual patch, I just removed the else if branch. -- Nx / talk 13:29, 1 February 2010 (UTC)

Done and appears to work fine. -- Daveh 02:47, 21 February 2010 (UTC)

Job QueueEdit

17 months on from this, I think the job run rate needs tweaking again.

There has been a lot of template editing recently, but that still doesn't explain what has been happening to the job queue length. As probably the #1 culprit for template edits, I've been keeping an eye on the length of the job queue and tonight it suddenly jumped from ~7,800 to ~14,500 when the only template that had been edited was one only used on a few pages.

In general, the job queue has borne no relationship to the amount of template editing being done. It soared after a few templates were edited and has generally dropped since then, but it seems as if some edits aren't triggering queue updates. When I simplified the {{NPC Summary}} template, we got over 1100 wanted links to Morrowind:Aggression as a side effect (irritating, but they'll go in the near future - hopefully), but only 270 links to Oblivion:Alarm (also unwanted - but there should be nearly as many as for MW).

To my - admittedly, only semi-tutored eye - it seems as if the job queue is struggling to keep up. The original link I provided referred to a server setup that simply couldn't cope with the strain, but a glance at the server status pages shows the current setup bearing the load well.

In other words, can I ask you to consider pushing up the $wgJobRunRate value a bit? I'll leave the final value to you, but it looks like a value of about 0.1 would be more sensible given the new server setup. –rpeh TCE 01:03, 11 January 2010 (UTC)

Yes, I've been thinking the same thing lately and was meaning to increase it a bit. We'll see if it changes anything in the job queue. -- Daveh 01:24, 11 January 2010 (UTC)
I've been watching the job queue and it looks like the Lore pages are causing it to rise. It rose regularly while Uniblab was editing but now that he or she has stopped it's going down. ‒ JoramTalk 05:59, 11 January 2010 (UTC)
If it keeps up querying the jobs table could help figure out the problem, for example SELECT job_cmd, job_namespace, job_title, COUNT(*) AS JobCount FROM job GROUP BY job_cmd, job_namespace, job_title might point to the cause. ‒ JoramTalk 21:12, 11 January 2010 (UTC)
Good idea...8,674 unique rows, job queue size of 16,044. Top rows include:
job_cmd job_namespace job_title JobCount
refreshLinks Morrowind Ascadian_Isles 147
refreshLinks Morrowind Ashlands 140
refreshLinks Morrowind Quests_in_Ald'ruhn 96
refreshLinks Tes3Mod Tamriel_Rebuilt/Quests 71
refreshLinks Morrowind Quests_in_Balmora 70
refreshLinks Oblivion 100%_Completion 53
refreshLinks Morrowind Quests_in_Sadrith_Mora 40
refreshLinks Morrowind Quests_in_Ebonheart 39
htmlCacheUpdate Oblivion Helping_Hands 38
refreshLinks Morrowind Quests_in_Vivec_Foreign 38
refreshLinks Morrowind West_Gash 38
refreshLinks2 Oblivion Helping_Hands 38
refreshLinks Morrowind House_Redoran 38
refreshLinks Morrowind Redoran_Quests 37
Daveh 23:57, 11 January 2010 (UTC)

Update: Changed the job run rate from 0.01 to 0.1. This should result in the job queue being completed 10x faster than it previously was. -- Daveh 00:03, 12 January 2010 (UTC)

I'm guessing all those came from recent updates to {{Quest Link}}. We'll probably hit most of 'em again tomorrow when {{Faction}} gets updates too. –rpeh TCE 00:08, 12 January 2010 (UTC)
Well the change has definitely had an effect... the queue is now at 28,450 and rising rapidly! –rpeh TCE 02:04, 12 January 2010 (UTC)
Mediawiki 1.15 Release Notes: Fixed excessive job queue utilisation No details are given so no way to know if it's related to current queue issues. ‒ JoramTalk 03:35, 12 January 2010 (UTC)
The queue is coming down now, about half what it was at rpeh's post. A programmer at MediaWiki also suggested a simpler version of the query, SELECT job_cmd,COUNT(*) from job GROUP BY job_cmd;. It'll show if there are more than just refreshLinks jobs causing problems. He also suggested using /maintenance/runJobs.php as a cron job but assumed this was a small wiki which I don't think it is at 1M views/day in June 2009. ‒ JoramTalk 06:52, 12 January 2010 (UTC)
That bugfix in MW 1.15 refers to this revision. If I understand it correctly, it's not a magic make everything faster patch, since it's mostly related to templates that have few backlinks, and the job queue usually gets filled when someone edits a high-use template. (And of course I would not recommend just applying this patch to your MW1.14, since it's fairly substantial and there may be bugs in it that have been fixed later, before the release of MW1.15) -- Nx / talk 09:19, 12 January 2010 (UTC)

(u/d) It's finally bottomed out at 4. It might be worth checking to see what those last four jobs are but otherwise, the change to the run rate has definitely been a good idea. –rpeh TCE 14:18, 12 January 2010 (UTC)

In fact it just hit zero. Definitely a good idea! –rpeh TCE 14:53, 12 January 2010 (UTC)

summoning from the void?Edit

moved to Shivering talk:Hunger#summoning from the void? --GKtalk2me 21:08, 2 February 2010 (UTC)

Icon ExtensionEdit

Unless you're planning an upgrade to MW 1.15 soon which eliminates the need for the Icon extension, when you have a moment can you upgrade Icon.php to the 1.6.2 version found at User:Joram/Icon.php (minus the <pre> tags of course). Version 1.5 that's installed now is bugged and the entire page will appear blank if the image isn't found. This makes the page a bit harder to edit if you save it that way. ;) They fixed that issue in 1.6.1 but borked the initialization code so it'll break any wiki it's installed on I think. I don't know PHP but it's close enough to C that I was able to fix it just by comparing with 1.5. I've tested 1.6.2 on my test wiki and it works fine though I think you know PHP so a second look would be good before you install. ‒ JoramTalk 00:13, 17 February 2010 (UTC)

I'll look into it when I'm back home and have some time. Will have to try it on a temporary copy of the UESPWiki to ensure all of our custom modifications don't mess things up. -- Daveh 21:47, 17 February 2010 (UTC)
Done! -- Daveh 02:44, 21 February 2010 (UTC)
Great, thanks Daveh. It seems to work perfectly as this did not break this page in preview: {{#icon:ThisImageDoesntExist.jpg}}. ‒ JoramTalk 21:48, 21 February 2010 (UTC)

Gadgets ExtensionEdit

While we're talking about extensions... please can you install Gadgets? It got a good response when discussed on the CP recently and would help out quite a bit for those without Javascript-fu. rpeh •TCE 01:01, 17 February 2010 (UTC)

Sure thing (when I'm back home and have a minute free as usual). -- Daveh 21:48, 17 February 2010 (UTC)
Done, but see notes on CP. -- Daveh 02:37, 21 February 2010 (UTC)

Performance Problems since the UpgradesEdit

Over the past couple of days I've been getting very erratic performance from UESP. For a minute or so, everything will be fine and then pages that normally load in a couple of seconds suddenly start taking 20-30s to load. This lasts another minute or so then everything is fine again. When the loading delay hits, all page loads are affected and then they will all clear within a couple of seconds of each other.

At first I thought it was my connection, but I've had absolutely no problems with other websites, so that doesn't seem likely. I did the basic network things like flushing my DNS cache, restarting etc etc with no luck. I get the same problem on both my computers. I've tried loading directly from content1 and content2 rather than going through www but both servers give the same problem. None of these problems occur when I'm viewing the maps - only when I'm looking at the wiki.

I've talked to a couple of other people on IRC and they don't seem to get quite the same problem although "sometimes it's fast, sometimes it's slow" was one comment. I don't see anything in the basic server status pages, but I can't access the Zabbix page as it asks for a username and password and my wiki login doesn't work.

The long and the short is that I don't think it's my computers or my connection and that it seems to be a wiki problem. Is there any chance something you've changed in the last couple of days is causing this? rpeh •TCE 10:59, 25 February 2010 (UTC)

Yes, I've been seeing the same thing about once an hour and since I couldn't see anything on the UESP server side I was thinking it might be my connection. I'm still trying to track down the source as there is nothing evident in the status/monitoring to indicate an issue (no load spikes, traffic issues, etc...). My guess right now is that randomly (1 in 10,000 say) a single page request gets caught up somewhere and takes a while to complete.
For now if you, and other admins, could just record the rough time and which page you were viewing when this occurs it will help in tracking it down. -- Daveh 13:38, 25 February 2010 (UTC)
It just happened on Recent Changes - of all pages - about 5 minutes ago. Exactly the same symptoms - no other page would load until that one finished. I found one or two other interesting things though. First, the HTML had all downloaded (I could do a View Source) so it must have been images, CSS or Javascript - and yes, I'm using the Gadget UTCclock that was also recently installed. Second, this was happening in Firefox so I flipped over to a copy of Opera I happened to have running, and I could reload Recent Changes in that with no problem. I'm going to switch to Opera for a while in case it's a browser problem, but that seems really, really unlikely. rpeh •TCE 15:00, 25 February 2010 (UTC)
Were you logged into the Wiki with Opera? I was running Firefox 3.5.8 when I encountered the issues too. I've seen the issue happen when it lags on getting the content as well as when getting a JS/CSS file (I think, I've only seen it a couple of times). -- Daveh 15:31, 25 February 2010 (UTC)
Additional...Note that for looking at the Zabbix monitoring you don't need to log in (at least you shouldn't have to). Just use the menu/links in the upper left. This is the same as logging in with the "guest" user, no password. If you look at the RecentChanges Load Time Graph the largest load time registered there is under 1sec in the past day or two (updates every minute). -- Daveh 15:35, 25 February 2010 (UTC)
Yes, I was logged in with both browsers (FF 3.6, Opera 10.10). I don't think the load time will show the problem: as I said, my browser definitely had the HTML so the monitor will probably call that a full load; it seems to be the images or other items downloading that causes the slowdown. It's not the gadget either - I tried turning it off and the same thing happened again. rpeh •TCE 16:02, 25 February 2010 (UTC)
I'm having this problem too every time I use the refresh button (forcing the browser to redownload js/css files). I checked with Chrome's debugging tools and Firebug, and I'm getting almost exactly 30s and sometimes even 1 minute latency on loading the files, but it's random, sometimes only one file is affected, sometimes all of them. Looks like something is hitting a timeout somewhere. I recall having a similar problem on RationalWiki (after a crash I think), which I fixed by restarting memcached. You said you reset memcached on the Community portal a few days ago, maybe that's the source of your problems? -- Nx / talk 17:35, 25 February 2010 (UTC)
I also just had an unusually long lag posting on rpeh's talk page. The page displayed but took about 40-45 seconds to finish loading. I'm using Opera Chrome so I can't tell what it's waiting for (or if I can I don't know how, it normally just says "waiting on"). Since I had UTCLiveClock functioning via JS before it was installed as a Gadget, I can confirm that that's unlikely to be the culprit. ‒ JoramTalk 17:51, 25 February 2010 (UTC)
Opera: Tools->Advanced->Developer tools, Network tab. Chrome: Page menu->Developer->Developer tools, Resources tab. -- Nx / talk 18:11, 25 February 2010 (UTC)
I was just discovering that page after you mentioned it.. I knew about the Developer Tools in Chrome but hadn't explored them yet. Nice to know they're in Opera too tho, I only use that for testing usually. ‒ JoramTalk 18:27, 25 February 2010 (UTC)

At least I'm not imagining it! It definitely seems to be CSS/JS. I'm surfing with images turned off in FF and still get the problem. Nx is right - it happens every time I hit the refresh button. rpeh •TCE 18:03, 25 February 2010 (UTC)
Try not using refresh, e.g. just hit the recent changes link on the sidebar to refresh recent changes. -- Nx / talk 18:11, 25 February 2010 (UTC)

Outdent Thanks to everyone's excellent tips it is indeed something related to the reload:

  • Reloading when logged in triggers it.
  • Reloading when logged out does not trigger it.
  • Delay is in multiple of ~30 secs.
  • It seems that randomly 10-12 items at a time can be downloaded. Any remaining items get delayed 30/60 seconds.
  • Only Wiki CSS/JS seem to be be affected. Having more than 10 images on a page doesn't seem to trigger it.
  • Some quick changes to some squid settings had little to no effect (might have increased the number of allowed objects from 10 to 12).

It definitely seems to be related to the number of CSS/JS items being downloaded at once and hitting a per user connection cap and timeout. None of the recent changes should have touched the CSS/JS (although that does remind me of one thing to check). I can play with settings and revert changes on content3 to test exactly where/when this bug was introduced. -- Daveh 20:02, 25 February 2010 (UTC)

It happens when logged out too, although it's a bit less likely, since there are less files to load. -- Nx / talk 21:00, 25 February 2010 (UTC)
I have been having all the same problems as Rpeh described them. Nx's suggestion works too I realized that yesterday it would load normally when you don't hit refresh, just the link on the sidebar, it will load like normal.--Corevette789 20:31, 25 February 2010 (UTC)
I've poked around a bit and restarted the memcached server and now the problem only shows itself on content1 not content2/3 (it's not a squid or db issue). This is strange as content1/2/3 are all configured almost identically, or should be at least. The stranger thing is when you reload a page on content1 and get the lag content2 doesn't respond but content3 is fine which makes no sense at all. At least it gives me something to do this weekend. -- Daveh 02:33, 26 February 2010 (UTC)
Well, it seems I broke my FireFox won't load any UESP Wiki page anymore. Restarting it, disabling add-ins has no effect. I'm using FireFox 3.5.8 on Windows XP Professional. Opera/IE still work with the random reloading issue. -- Daveh 03:42, 26 February 2010 (UTC)
It seems this was two issues. The FireFox issue was unrelated and has been seen by a few people (clearing the entire history and restarting seemed to fix it). The site reload issue is a separate beast. -- Daveh 03:50, 26 February 2010 (UTC)

(u/d) This issue is still ongoing so it's not related to the hard disc problem. Talking to people on IRC shows that one or two people will have problems while others are fine and then the situation will reverse a few minutes later. Personally, the site has gone between fine and unusable over the course of the day - often minute to minute. I know you're looking into it, but things are getting worse rather than better. rpeh •TCE 22:34, 28 February 2010 (UTC)

I've swapped content3 for content1 in the Squid cache so you shouldn't see any issues from since only content1 seems to be having the "reload" issues. If, however, you do notice any issues still just let me know as usual. -- Daveh 00:41, 1 March 2010 (UTC)
I've actually found and fixed the problem on content1 (old NFS server applications hanging around after the switch to files1) and have reverted the content1/3 switch on squid1. -- Daveh 01:26, 1 March 2010 (UTC)
It all looks great now - faster than ever! Thank you for sorting it all out. rpeh •TCE 02:01, 1 March 2010 (UTC)
Much better! Thanks for fixing it! --GKtalk2me 02:43, 1 March 2010 (UTC)


Hi Dave. I'd forgotten about the issue of having Wanted Categories still showing up from the database corruption a little while ago. Rpeh reminded me today that that was the cause of it. I just had a look at the MediaWiki query for Special:Wantedcategories and I believe the following query should give you the invalid rows left over from the database corruption.

SELECT * FROM categorylinks WHERE cl_from NOT IN (SELECT page_id FROM page)

Assuming that returns 5 rows for the 4 invalid categories on Wanted Categories (Oblivion-Places-Imperial Guardtowers is legit, so it shouldn't appear), then you should be able to just delete them at that point. I'm still familiarizing myself with the table structure of MW, so there could be more to it than the SELECT statement above, but it's a pretty simple table, so I'd be really surprised. Obviously, you'll want to back up the wiki data before deleting anything. Robin HoodTalk 05:59, 19 March 2010 (UTC)

That query actually returns a bunch of rows:
  cl_from       cl_to   cl_sortkey      cl_timestamp
  7544   Marked_for_Deletion            Sanguine/Description    2009-12-11 15:59:55
  7544   Oblivion-Quest_Descriptions    Sanguine                2007-07-03 21:45:47
  7544   Proposed_Deletion_(Bot)        20091211205954          2009-12-11 15:59:55
  7566   Lore-Books-Descriptions        Advances in Lock Picking        2008-07-05 08:06:36
  7566   Marked_for_Deletion            Advances in Lock Picking/Description    2009-12-14 11:48:07
  7566   Proposed_Deletion_(Bot)        20091214164807          2009-12-14 11:48:07
  7733   Marked_for_Deletion            Garlas Agea/Description 2009-12-11 15:15:55
  7733   Oblivion-Place_Descriptions    Garlas Agea             2008-01-04 06:12:14
  7733   Proposed_Deletion_(Bot)        20091211201555          2009-12-11 15:15:55
  11294  Marked_for_Deletion            Books/Biography of Barenziah v II       2009-08-12 17:06:33
  11294  Proposed_Deletion_(Bot)        20090812210633          2009-08-12 17:06:33
  11294  Redirects_from_Alternate_Names Biography of Barenziah, v 2     2007-10-26 22:52:22
  46049  Marked_for_Deletion            Hassour/Description     2009-12-13 10:45:25
  46049  Morrowind-Place_Descriptions   Hassour                 2009-11-29 12:53:36
  46049  Proposed_Deletion_(Bot)        20091213154525          2009-12-13 10:45:25
  60882  Marked_for_Deletion            Telaendril's Ocheeva Note/Author        2009-12-14 11:25:43
  60882  Oblivion-Books-Authors         Telaendril's Ocheeva Note/Author        2008-08-29 06:10:09
  60882  Proposed_Deletion_(Bot)        20091214162543          2009-12-14 11:25:43
  60887  Marked_for_Deletion            Jearl's Orders/Author   2009-12-14 11:10:04
  60887  Oblivion-Books-Authors         Jearl's Orders/Author   2008-08-29 06:13:02
  60887  Proposed_Deletion_(Bot)        20091214161004          2009-12-14 11:10:04
  66581  Marked_for_Deletion            Tamriel Rebuilt/Black Roses for Banviris/Descripti...   2009-12-11 20:38:58
  66581  Proposed_Deletion_(Bot)        20091212013858          2009-12-11 20:38:58
  66581  Tes3Mod-Tamriel_Rebuilt-Quest_Descriptions     Black Roses for Banviris        2009-02-16 16:34:47
  66802  Marked_for_Deletion            Tamriel Rebuilt/The Ring of Ineptitude/Description      2009-12-14 10:27:28
  66802  Proposed_Deletion_(Bot)        20091214152728          2009-12-14 10:27:28
  66802  Tes3Mod-Tamriel_Rebuilt-Quest_Descriptions     Ring of Ineptitude, The         2009-02-16 16:35:45
  68926  Marked_for_Deletion            Tamriel Rebuilt/Hunted, Volume III/Author       2009-12-14 10:15:34
  68926  Proposed_Deletion_(Bot)        20091214151534          2009-12-14 10:15:34
  68926  Tes3Mod-Tamriel_Rebuilt-Books-Authors  Hunted, Volume III      2009-03-08 14:27:44
Note that the query was actually run on content3 since db1 is still running MySQL v4.0 which doesn't support sub-queries (you could probably rewrite it as a join). The two versions should be the same although I will confirm that first. I can easily delete/move these rows if needed (I'll look into a bit more just to make sure it won't break anything obvious). -- Daveh 01:39, 22 March 2010 (UTC)
Hahaha...I deliberately wrote it as a nested query because I'm not very familiar with MySQL (almost all my experience is with SQL Server and Access) and wasn't sure how well it would deal with a join if you just changed the SELECT to DELETE had it returned what we were expecting. The equivalent join query would be:
SELECT * FROM categorylinks LEFT JOIN page ON cl_from = page_id WHERE page_id IS NULL
It looks like it includes the problem rows nevertheless (the ones where cl_to matches a page in the Wanted Categories):
  66581  Tes3Mod-Tamriel_Rebuilt-Quest_Descriptions     Black Roses for Banviris        2009-02-16 16:34:47
  66802  Tes3Mod-Tamriel_Rebuilt-Quest_Descriptions     Ring of Ineptitude, The         2009-02-16 16:35:45
  46049  Morrowind-Place_Descriptions   Hassour                 2009-11-29 12:53:36
  7733   Oblivion-Place_Descriptions    Garlas Agea             2008-01-04 06:12:14
  68926  Tes3Mod-Tamriel_Rebuilt-Books-Authors  Hunted, Volume III      2009-03-08 14:27:44
Try this query, which is just an ungrouped version of the one they use for Wanted Categories, and let's see if that gives us any insight. It should just return the above set of rows plus the legitimate one, but at least then we'll be sure which rows we need to deal with.
SELECT categorylinks.* FROM categorylinks LEFT JOIN page ON cl_to = page_title AND page_namespace = 14 WHERE page_title IS NULL
Robin HoodTalk 08:20, 22 March 2010 (UTC)
I was thinking about it earlier, and I think I understand now why all those rows appeared: they're legit...or rather, they're legitimately incorrect rows! The reason they don't show on Wanted Categories is because the categories exist, but in reality, those category links have no matching pages. If we were to, for example, clean out "Proposed Deletion (Bot)" and then delete the category, the above rows would show in the Wanted Categories list, but have no traceable page that they were coming from. You'll want to try it on a test server just to be certain, but I'm reasonably certain that all the rows from the initial query should be deleted. Robin HoodTalk 23:04, 22 March 2010 (UTC)

New User GroupEdit

I'm jumping the gun slightly here, but it's necessary. You may have seen this discussion about extending the "block" right to other users. There has been a huge upswing in vandalism in the last few days and often, our two active admins have been busy with their own lives and unable to respond for some time. The consensus looks like being that there should be some other way of granting block rights to users. Since this is something only you can set up, and you're often busy, I thought it would make sense to ask you asap and then the details can be hammered out later.

So, please can you create a new "Block" group that can be assigned by members of the sysop group (like Patrollers can be now?) Since it's only one right being discussed, that provides enough granularity to give it to all patrollers, individual patrollers, or anybody else as and when necessary - so it should cover all possible bases. rpeh •TCE 22:35, 19 March 2010 (UTC)

Seconded. We're still discussing exactly how it'll be done, but it would be nice to have the group already created so we can grant the rights as we deem necessary. --GKtalk2me 23:06, 20 March 2010 (UTC)
Created a "blockuser" group with only the "block" permission set. Let me know if you need anything else. -- Daveh 01:26, 22 March 2010 (UTC)
I see the group in User rights management, but it's in the column of "Groups you cannot change". I suppose the Administrators group rights will have to be adjusted to allow admins to grant the right. I also wanted to make sure that the blockuser group will have the option to prevent account creation and the option to automatically block the last IP, as well as the option to block only anonymous users when setting a block. The options that shouldn't be available to those with only blockuser rights are the options to block a user from sending emails, and to block a user from editing their own talk page. --GKtalk2me 15:34, 22 March 2010 (UTC)
The "blockuser" group just has the "block" right and whatever that entails. I'll have to research exactly what rights to grant/deny based on your requirements. I'll also look into seeing how to make the group usable/settable by the current admins and not only sysops (just me at moment) or if that is even possible. -- Daveh 16:35, 22 March 2010 (UTC)
Those extra rights should be included in "block", so there's no problem there. The "Patroller" group is assignable by admins, but I'm not sure if that was something included in one of Nephele's custom updates or not. rpeh •TCE 17:02, 22 March 2010 (UTC)

(Update)Nx has posted the extension, and has made a few comments related to the discussion: [1] --GKtalk2me 18:48, 22 March 2010 (UTC)

Any new news on how/if the group can be assignable by admins? --GKtalk2me 18:51, 26 April 2010 (UTC)

Image IssuesEdit

We seem to have another image caching issue. What follows doesn't just come from my own experience but from others too. There's nothing too important so far, but it bears mentioning. The following issues have occurred:

When uploading an image the result screen is a "No image of this name exists" form. After uploading an image, the preview is displayed with the old image but with the new image's dimensions, resulting in a scaled image. When viewing a page containing a new image, a red link to the new image is displayed rather than a thumbnail of the new image.

In my own experience, waiting a while (15 mins) and reloading solves the problem, but there was a case the other night where File:OB-npc-Neesha.jpg had not updated after some hours. In this case, as with others, doing a purge and/or a hard refresh fixed the problem, but it seems there is an underlying problem.

Please could you take a look at the image caching? Thanks. rpeh •TCE 01:51, 23 March 2010 (UTC)

"After uploading an image, the preview is displayed with the old image but with the new image's dimensions, resulting in a scaled image" This is normal, it's because of your browsers cache (i.e. not much can be done server-side). The other problems are not, of course. -- Nx / talk 10:01, 23 March 2010 (UTC)

MW 1.14.1Edit

Would you be willing to upgrade the site to MW 1.14.1? It would solve a small problem we're having with hover text not appearing on some icons. While it's obviously a very minor problem, Nx assures me that it's also a very minor solution (see this discussion). Thanks either way! Robin HoodTalk 02:59, 25 March 2010 (UTC)

Sure, I'll take a look. -- Daveh 11:32, 25 March 2010 (UTC)
Unfortunately it doesn't fix the problem, since the bug is only fixed in MW 1.16. -- Nx / talk 11:39, 25 March 2010 (UTC)

New AdminEdit

Krusty has been approved by the community and it's been about a week, so can you please give him the usual rights? Thanks! --GKtalk2me 17:19, 3 May 2010 (UTC)

Done! -- Daveh 17:22, 3 May 2010 (UTC)


Hi Daveh. I'm Caz, a moderator on the UESP forums. Codex has sent you a pm which needs your attention as soon as possible. It's rather important. Thanks! Caz 21:55, 4 May 2010 (UTC)

I just got back from a short trip and will take a look after work today. -- Daveh 13:11, 5 May 2010 (UTC)
Thanks a lot :) Caz 15:19, 6 May 2010 (UTC)
There's another pm waiting for you about this matter, we'd appreciate it if you could check it soon. Thanks in advance. Caz 12:16, 6 June 2010 (UTC)

Site Slow - skins/images problem?Edit

I'm getting really poor performance from the wiki at the moment. Most pages start to load fine but then halt waiting for or (ie, the same server). When logged out that doesn't happen, presumably because of the caching. The server-status page is currently showing 318 connections but content1's status shows only 2. Could you take a look when you have a minute, and maybe give that server a kick? Thanks. rpeh •TCE 18:21, 16 May 2010 (UTC)

Yah, I was noticing it too on the weekend and tried a few things including deleting the 97GB (!) lighttpd log file on files1 and restarting the web server. It didn't have an immediate effect but it slowly improved in the afternoon. I'll keep an eye on it in case there are other issues lurking about. -- Daveh 13:50, 17 May 2010 (UTC)
Thanks. It's definitely been a lot better today. There are still over 200 connections though, which seems a little high to me. rpeh •TCE 16:24, 17 May 2010 (UTC)
It's doing it again. Is it just a Sunday problem? rpeh •TCE 19:21, 23 May 2010 (UTC)
It happens a lot on the weekends to me, it does say it is waiting for skins too. I have slower internet so I think it might be happening more to people with DSL. Another thing: Sometimes the "skins: do not load at all and I get a HTML looking screen. It's just white and the text is considerably bigger.--Corevette789 19:28, 23 May 2010 (UTC)
Update -- I disabled the lighttpd access log on files1 (image/skin server) and restarted it. I also increased the file handles from 1024 to 8192 although I doubt this would have been the source of the issue. This appears to have solved the slowness issue for the moment anyways. If you see it crop up again just let me know. -- Daveh 20:06, 30 May 2010 (UTC)
I also rooted around the mysql logs yesterday, and noticed some issues being caused by MetaTemplate -- in particular there was a database deadlock on May 23, which I suspect was at least part of the problem last weekend. I've made some code changes to address the issues, and I've just been finishing up the debugging. I'll probably try to upload the new code sooner rather than later (sooner means more risk of new bugs surfacing while the site is busy; later means more risk of another deadlock, which is probably the greater risk). --NepheleTalk 22:52, 30 May 2010 (UTC)
It's starting to do it again. Sometimes pages are fine but more often I'm seeing the delay waiting for skins.uesp again. rpeh •TCE 14:55, 5 June 2010 (UTC)
I realized that most of the time, users shouldn't even need to access skins.uesp -- that content is all static and shouldn't be reloading every time you view a page. After rooting around, I found that expires headers weren't being set on any of the lighttpd content. So I turned on expire_mod and set all files to expire after 1 month (which was the expiration being used for these files when content1/2 were the servers). I'm guessing that should make a noticeable difference... but we'll have to wait and see. --NepheleTalk 06:03, 9 June 2010 (UTC)

Additional Note for Lost Amusei in the Ultimate HeistEdit

Discussion moved to Oblivion talk:The Ultimate Heist#Additional Note for Lost Amusei in the Ultimate Heist

content3 groupsEdit

Hi, Daveh! I just noticed that I'm not part of the uesp group on content3, which means I have no ability to update the code for w/extensions/MetaTemplate on that server (conversely, I am part of the uespadmin group, which isn't the case on content1, although that's not really a problem AFAIK). If you're around, could you quickly add me to the uesp group? Or else update the contents of w/extensions/MetaTemplate so it matches what's on content1 and content2? Thanks! --NepheleTalk 18:12, 31 May 2010 (UTC)

Oops, one more issue: I don't think I have any access to I made a tweak to prefs.js but I can't put it in place (FYI, the new prefs.js is at w/skins/common/prefs.js on both content1/content2, if it's easiest in the short term for you to just copy that one file over). Thanks again. --NepheleTalk 05:00, 1 June 2010 (UTC)
OK, false alarm :) I've been able to take care of the issues myself (I remembered that I'm on iWeb's list of special people...). --NepheleTalk 18:55, 3 June 2010 (UTC)

Virus found on siteEdit

Discussion moved to UESPWiki:Administrator Noticeboard#Virus found on site

Major Site ProblemsEdit

I'm having some major site problems currently. Reading pages seems to be fine, but Show Preview and Save Page almost invariably fail...maybe working 1 time in 20. Looking at the server statuses, it looks like Content3 has more backlogged activity than usual, and Content2 has an inordinately high CPU load compared to what I'm used to seeing. I don't know if those are actually meaningful observations, though. Robin HoodTalk 18:02, 9 June 2010 (UTC)

Just to add to this, the error I get back most often is "Zero Sized Reply", which the first Google hit I got indicated was a Squid error. It seems Krusty had no problems performing an edit for me, however, so I'm not sure what to think at this point. Also, some things, like the above message and this one, seem to do just fine on Show Preview and Save Page; most things don't, however. Robin HoodTalk 19:32, 9 June 2010 (UTC)

Arny, no this is a user's Talk page. RH, Everything is fine for me on all 3 servers. Well, now at least.--Corevette789 19:37, 9 June 2010 (UTC)
I'm not seeing anything unusual right now, and the apache logs on content1/2 don't contain any unusual messages from the last few hours. (At first I thought a batch of unusual messages about -- "Connection refused: proxy connect to port 80 failed" -- might be associated with this, but when I checked the times of those messages they're all from last night, and correspond with the time when I restarted lighttpd on
On the other hand, the CPU usage on content2 does seem high -- even taking into account that I restarted its apache last night (there's always an initial CPU spike after a restart). It's definitely a lot higher than content1, which was restarted at the same time. I'm not sure what would be responsible for that. --NepheleTalk 19:51, 9 June 2010 (UTC)
I haven't noticed this- but I've been doing other things off-wiki for a while, however making this edit I didn't see anything more than the normal lag for 6 word documents and 3 tabs open.--TheAlbinoOrc 20:00, 9 June 2010 (UTC)
Actually saving did take a while but show preview worked fine.--TheAlbinoOrc 20:01, 9 June 2010 (UTC)
(edit conflict) It's not just UESP, it's affecting me on Wikipedia as well, and it seems like it's large wiki submissions that are getting borked. I may have just figured out what was causing the problem, which was an unrelated router change I made yesterday. Why it only became a problem today, though, I don't know. In any event, I'm gonna save now while the saving's good and see if I can get back to work on other things or if this apparent burst of functionality is just a fluke. Sorry to have troubled you, Nephele! Robin HoodTalk 20:11, 9 June 2010 (UTC)
(edit conflict) I'm thinking the high CPU usage on content2 is because memcached is completely disabled there. I can see in the upgrade history why Daveh did that, but there are several problems with the current configuration -- we've got all the problems that a shared memcached was supposed to fix, plus absolutely no memcaching on two-thirds of our servers. I'll need to think about / research how to configure memcached some more before doing anything, however. --NepheleTalk 20:12, 9 June 2010 (UTC)
re: Memcached -- I disabled it as it was actually slower with memcached on content2/3 than without, probably due to the servers beings on different networks. iWeb was charging a minimum of 400$/month to move servers to the same rack although I see now there is a cheaper option at 175$/month that we could probably manage cost wise. This would solve a variety of minor issues and is really what we should be doing anyways, especially if we're looking at increased site traffic (which may be the case if ES5 or a ES-MMO is in the near future). -- Daveh 21:40, 9 June 2010 (UTC)
Hi, Daveh! We've been keeping your talk page busy ;) I saw that comment about memcached on the upgrade history. My first thought is that just going back to the old setup -- with local memcached for each of content2/3 -- would probably be better than what we have right now. We're back to the situation where memcached can be out-of-sync, since each server is using a different copy of memcached. But we're effectively already back to that situation -- content1's memcached isn't being kept up-to-date if content2/3 handle an edit. --NepheleTalk 21:59, 9 June 2010 (UTC)
Ahh...that's true. I had been doing some simple benchmarking a while ago but got sidetracked and never came back to it. Its a simple matter to get memcached running separately on each content server and see how it works. -- Daveh 01:34, 10 June 2010 (UTC)

() Yup, clearly everything's working now and it was related to that router setting (which should have had no effect AFAIK). Either that feature of our router is buggy or my lack of knowledge of it led me to configure it incorrectly. But hey, at least I discovered the Content2 issues while I was having problems. :) Robin HoodTalk 20:20, 9 June 2010 (UTC)

Okay this is odd, I had the connection time out a few minutes ago but it wasn't me because the other tabs worked fine, anyone else have this happen?--TheAlbinoOrc 21:10, 9 June 2010 (UTC)
Yeah, something just happened a few minutes ago and the site got slow for everyone. Robin HoodTalk 21:19, 9 June 2010 (UTC)
It happens the same time of day for everyone. This has been going on for quite a while IIRC.--Corevette789 21:21, 9 June 2010 (UTC)
I'm convinced the "regular" slowdown is an issue with squid1. I finally managed to get logged onto squid1, and can see one problem right away ("WARNING! Your cache is running out of filedescriptors" in the logs), plus a lot of other unhappy types of log messages. I've got to run right now, but tonight I'll try to spend some time fumbling around with squid configuration -- see what's necessary to fix the obvious error messages, and also try to explore whether there's anything specific that the squid is being told to do every day that might be responsible for a recurring slowdown. Beyond that, it's also possible that we're just maxing out the capacity of that particular server.... --NepheleTalk 21:52, 9 June 2010 (UTC)
I'm having slowdown right now so if it's helpfull to be able to check this while it's going on here's one. (No it isn't me I shut everything down and it still took 17-20 seconds just to load my watchlist)--TheAlbinoOrc 19:35, 10 June 2010 (UTC)
It also took almost 40 seconds just to save my previous edit.--TheAlbinoOrc 19:36, 10 June 2010 (UTC)
Really slow here, too. The Server Status pages show slightly more traffic than usual but nothing too OTT. The only oddity I can see is that content3 is showing a lot of "waiting for connection"s rather than "open slot"s. rpeh •TCE 19:41, 10 June 2010 (UTC)
I've increased the file descriptors on squid1 from 1024 to 8192 (4096 usable by the Squid cache) and will see if we still get the error messages. I'll also take a quick look on the other servers and ensure the limits are increased (a few previous issues were caused by the default number being on the low side). -- Daveh 01:30, 11 June 2010 (UTC)
Oops, I already changed it from 1024 to 4096 last night. We haven't had any out-of-file-descriptors errors since then, but we still did get slowdowns today. The slowdowns do seem to be associated with squid1 handling large numbers of connections (250+ maybe), but I couldn't see what exactly was maxing out to cause a slowdown. --NepheleTalk 01:40, 11 June 2010 (UTC)
Okay this is odd. Around 5:30 EST there was major site slowdown I would have posted about this then but trying to save the change took over 100 seconds the first time. I shut the site down but it wouldn't even load for a while. Anyway the oddness is- If it's related to large numbers of conections shouldn't it happen at roughly the same time each day not have a couple of hours difference?--TheAlbinoOrcGot_a_question? 22:59, 11 June 2010 (UTC)
Content2 crashed. See Upgrade History. Robin Hoodtalk 23:03, 11 June 2010 (UTC)
Ah, sorry I hadn't realized that.--TheAlbinoOrcGot_a_question? 23:05, 11 June 2010 (UTC)
I was unable to connect to the site without using the content servers for about half an hour after that for about 30 minutes I was unable to access the content servers or the main site.--TheAlbinoOrcGot_a_question? 21:12, 14 June 2010 (UTC)
I restarted content2. As far as I can tell, the problem is basically just that we're getting spikes of activity which are more than our servers can handle. I know that when any of our content servers get too many requests at once, they basically freeze up (I saw it happen with content2 a couple weeks ago; I've seen it happen in the past going back more than a year). When this happens, it's impossible to even login in to the server (probably apache has monopolized all the memory). If you wait long enough (half hour or more), apache starts to forcibly close timed-out connections and eventually sorts itself out, but it takes quite a while and even then, there seem to continue to be frozen connections for hours (maybe even days) -- probably until apache is restarted. However, I don't have time right now to investigate more -- for example, to see whether there are a couple of IP addresses responsible for most of the requests. --NepheleTalk 21:54, 14 June 2010 (UTC)
I've been looking into the issues when I have time but unfortunately the past few months have been more hectic than usual for me as will the next few months. Fortunately, the issues are not having a significant impact on the overall site traffic yet...for example last weekend had the highest traffic we've seen in a few months (only by 5% or so) so it is not like we're loosing half the traffic every time these issues pop up. Current issues as I see them are:
  • Squid1 is beginning to have very high IO load (up to 50% wait times in top) which I assume is due to unoptimized cache or hard drive/partition settings. My one attempt to fix it resulted in a temporarily dead server so I'm hesitant to mess with it again until I have dedicated time to deal with it.
  • Content2 randomly has very high loads for no apparent reason. Update: Actually as luck had it I left top running on content2 today and since it froze after the recent reboot the %wa was very high (90%) indicating some sort of IO issue.
  • Content3 is also randomly "spazzing" out but since it doesn't handle any real purpose at the moment (mainly backups which I have to check on) I'm not worrying about it too much.
-- Daveh 22:25, 14 June 2010 (UTC)

() (*wondering if he's guilty look*) Does this mean that we should avoid doing things like going down Recent Changes Ctrl-clicking on "diff" to open a bunch of pages at once? Robin Hoodtalk 22:20, 14 June 2010 (UTC)

Just jotting down a few things I'm noticing while looking through the squid logs (times should all be EDT):
  • When squid was restarted at 21:39, there were about 475 connections that had to be closed "due to lifetime timeout" -- that's a huge number of connections for squid to be handling at once even if they were all active connections; if any significant number of those connections were dead, it may suggest there's also a problem with not releasing dead connections.
    • During the previous restart at 2010/06/13 12:18 about 675 connections had to be closed due to lifetime timeout
    • When squid was restarted at 2010/06/09 22:27:15, only 1 connection had to be closed; at 2010/06/10 21:23:23, only 17 connections; at 2010/06/11 18:11:19, 425 connections
    • Did something done on ~June 10 cause a problem?
  • Squid made maybe 20,000 failed attempts to connect to content2 between 16:15 and 17:15 -- if each of those was because of an incoming squid request, that suggests 80,000 incoming requests during that hour (half of requests are typically dealt with by the cache; half of the requests go to content1). I have no idea whether that's typical, but it seems like an incredibly high number.
I'll do more followup later, but I wanted to post this info before heading out for a couple hours. --NepheleTalk 01:15, 15 June 2010 (UTC)
Thanks...good stuff. Note that we get on the order of 600k page views daily which probably works out to around 50k page views/hour during peak time. Add in the fact that each page view can technically be 10-20 separate file requests and 80k doesn't look so bad. It still could be some sort of flood of requests from somewhere, malicious or not. -- Daveh 02:37, 16 June 2010 (UTC)
Another thing I just noticed as traffic over the past few weeks has been slowly increasing and it is currently 20% than it was 2-3 weeks ago. The slight amount (an extra 100k page views/day) of traffic increase may have been enough to cause problems to show up that were not noticeable previously. I think the current server setup is more than enough to serve significantly more traffic but the countless server configurations are likely far from optimized. -- Daveh 02:45, 16 June 2010 (UTC)
Is it okay to edit if I'm on the contents? I wanted to check since content1 is running much faster than the main site right now.--TheAlbinoOrcGot_a_question? 23:35, 15 June 2010 (UTC)
I hope so I just checked my talk on conten1 and forgot completely that I wasn't on the main site (although I did revert my change)!--TheAlbinoOrcGot_a_question? 23:42, 15 June 2010 (UTC)
Yes, it is perfectly fine. You are only bypassing the Squid cache so technically it will be a bit slower but probably not at all noticeable. Note that this might not always be the case. There may be times where content1 is down on purpose. One of the benefits of using Squid, or a similar process, is that we can easily change which servers are serving the content. -- Daveh 02:37, 16 June 2010 (UTC)

(outdent) Content1 seems to be dead on its... racks? at the moment. I just got this server status page - content2 was fine at the same time. The site, as you might expect, was not going very quickly. rpeh •TCE 11:59, 21 June 2010 (UTC)

I was having pretty major site slowdown for a while (bad enough that I couldn't even use the main site in fact). Could this be related to the recent changes to content3?--TheAlbinoOrcGot_a_question? 21:50, 25 June 2010 (UTC)

Deleting Wiki Acount / Renaming Wiki Account...Edit

Hello, I was hoping you or possibly another Admin could assist me in either A) Renaming my account or B) Deleting it all together. I created the account a lonh time ago and never liked te name (At the time it was all I could think of.) I would like to have it changed to Pauper if possible. Thanks in advance, Brynn.Shadow Hide You 19:51, 14 June 2010 (UTC)

Done. --Krusty 20:00, 14 June 2010 (UTC)
The account still seems to be here. O.o Apperantly renaming didn't work, could you perhaps delete the account and i'll start over? Thanks, Brynn.
I'm not an admin, but looking at your user page and contributions, there's a bit of a mismatch — the user page says the account Pauper isn't registered, but I can see a list of Pauper's contributions! I think probably something was just missed, which may cause some issues. I would suggest just waiting for a bit for Krusty or one of the other admins to see what the problem is and hopefully fix it. Robin Hoodtalk 22:59, 14 June 2010 (UTC)
Just to find out what is going on - could you give your new username (Pauper) a shot and log in with that?

Says Pauper cannot be found, I was writing these messages from my iPhone which may have caused some problems, I logged off of that one and now it's just this computer. I can log in with Emer but Pauper doesn't exist.Shadow Hide You 23:03, 14 June 2010 (UTC)

Seems like we have a problem with the name-change feature. Hang on, though - hopefully Daveh or Nephele will be able to help out when they read this. --Krusty 23:13, 14 June 2010 (UTC)
Alright, thanks alot for the help! — Unsigned comment by Pauper (talkcontribs)
I just tried using the renameuser tools again, and it seems to have been more successful this time. Or at least, I can confirm that in the database, the user_name for user_id=21801 is now Pauper, not Emer -- which wasn't true before I reran renameuser. I'm not sure what would have happened differently the first time. It's possible that there's an obscure problem with our server configuration messing things up (namely that memcached isn't properly clearing the old user info, because our servers aren't sharing memcached the way they should be). I'll continue to check some things behind the scenes, but hopefully the account is now useable. --NepheleTalk 00:21, 15 June 2010 (UTC)

The account seems to be very glitched, I'm logged in but the second I change to a different page, I'm logged out.

Are you using a satellite internet provider (e.g, Hughes)? If not, try clearing all UESP cookies, then log in again. --NepheleTalk 01:17, 15 June 2010 (UTC)

Dave Humphrey's Furniture Mod (Green House Meshes)Edit

Dave, It's been a while since you've been around Morrowind, I'm sure. I have a question about your furniture mod. It came with a set of green house meshes. I wanted to know if I may use them in a mod I am making for Morrowind. Thanks,

Melchior Dahrk 01:16, 16 June 2010 (UTC)

Yup, of course. Have fun with them! -- Daveh 02:30, 16 June 2010 (UTC)


Question From A New UserEdit

Hi Daveh, I am new to editing stuff on this web site. I've only been active for a few days now and I am still trying to figure out how things work. Apparently I have caused a problem with patrollers by clogging up the site with multiple edits to things I post as I recently received a message from someone called Gum!!! about it. The problem occurs because of my wanting to say things just right. I look at stuff I've posted and I find better ways of saying it and go in and change it. I am using the preview button everytime but as I said I have a problem with wanting everything I say to be perfect. From now own I am going to do my best to refrain from this as it looks like it is causing a problem. A question that I have is about my user page. I am changing it as I progress through the game. Should I wait until I finish the entire game or a story line before adding something about my character or is it ok to add things to my user page as I go? If I add things as I go it might clog up the edits page and cause the problem mentioned above. Anyways any helpful feedback that you could give me would be appreciated because I don't want to cause problems on this site. Chris McCallum Let's Talk 15:27, 22 June 2010 (UTC)

You can see a discussion on this topic here. Basically, don't overdo it. The only person interested in your character gaining a rank in the thieves guild is you, so it's not something that needs a dozen posts in a day. It's fine to keep an up-to-date character sheet, but in general it should be updated infrequently. rpeh •TCE 15:45, 22 June 2010 (UTC)
A good idea for me would be instead of keeping a log on my user page for every character I play to change the log to things that I have accomplished in the game with any character and never edit it again once it has been posted to a section. Chris McCallum Let's Talk 16:19, 22 June 2010 (UTC)
The method you use to update is up to you. Everybody makes mistakes or re-reads something they just wrote and realizes it was poorly worded. As Patrollers, all we ask is that before you hit "Save", not only should you click "Show preview", but if you're a picky editor (first of all, welcome to the club, but more to the point...) re-read what you wrote at that time, and if necessary, sit on it for a few minutes and re-read it yet again so that you're less likely to have a multitude of edits to the same page. Robin Hoodtalk 18:42, 22 June 2010 (UTC)
Hi Robin Hood, just got your response to what has been going on with my editing. I didn't know that my editing over and over again was causing a problem for the patrollers of the site but now that I am aware of it you and the others can expect to see a lot less of it. I went ahead and changed my user pages achievements to I have done this instead of my current character has done this. Once I add an accomplishment that I have done in the game, I will never delete it or edit it again. There is no point in doing it for each character you play and besides that most people on this web site are not going to ever look at my user page anyhow. Anyways I thank you for your helpful feedback as I learn more and more about how to use this site properly. Chris McCallum Let's Talk 19:56, 22 June 2010 (UTC)
Sounds like a plan. Glad rpeh and I could help. Robin Hoodtalk 20:02, 22 June 2010 (UTC)


Hey Daveh,

I've been using UESP for a while now and I've finally got around to getting accounts on both the forums and the main site, however none of the verification emails are sending. Tried having it resend several times now and still nothing. Help me out man!

Thanks! =)

Talon Wulf 07:50, 29 June 2010 (UTC)

Is this you?Edit

Is it your Furniture Store mod that made it to the PES Hall of Fame for July? If so, congratulations! rpeh •TCE 20:35, 8 August 2010 (UTC)

First Edit EverEdit

Hi Dave (or Neph if she sees this first <g>). I just tried to find out when the first edit was made on the wiki, and was rather surprised when this link (query for oldid=1) returned an error message. Querying random really low numbers, I found that a great many were invalid, yet others weren't, seemingly at random. Did something happen that got rid of (or renumbered) a lot of edits at some point, or am I just incorrect in assuming that oldid=1 must be the first edit? I'm not really all that concerned about when the first edit was, I just wanted it for an example...I'm more worried about the possibility of corruption. What was amusing about this is that what appears to be the first edit ever, by MediaWiki default during the install process, currently has an oldid of 824, but the next edit has an id of 100. Robin Hoodtalk 21:37, 8 August 2010 (UTC)

Missing revision ids are because pages were deleted. Missing ids are nearly always found in the archive table listing revisions from deleted pages (as opposed to the non-deleted ones in the revision table). Even the few missing ids that I couldn't find in the archive table appear to be revisions that are actually listed there, but without an id (rev_id=NULL), probably because an early version of the wiki software handled revision ids differently. In short, I don't see evidence for any revisions going missing in the database (other than the handful that I already identified and dealt with when we lost a hard drive last summer).
Furthermore, a large fraction of the early wiki edits are on deleted pages -- some were test pages, but more importantly, hundreds of the first pages were default pages in the Mediawiki: namespace (i.e., all of the edits by "Mediawiki default"); nearly all of those pages (683 out of 727) were later deleted when the wiki software found a better way to deal with the default settings. In addition, revisions 98-824 by Mediawiki default all have an identical timestamp (including both deleted and non-deleted revisions), down to the second, one second being the resolution of the wiki's timestamps. The user contribution history is only sorted by time, so in typical database fashion, the order in which those simultaneous edits are displayed on Special:Contributions/MediaWiki default is effectively random.
That said, there is a bit of strangeness with the timestamps -- the Mediawiki default edits are all at 20050112023532, but there are lower number revisions (and revisions in the middle of the Mediawiki numbers) from February 2005. Maybe the software tried to reuse revision ids in early versions; more likely is that the revision-numbering was revamped by an upgrade (i.e., if revision numbers used to be page-specific instead of system-wide -- or if revision numbers didn't even exist, just timestamps and page numbers). Knowing how much the software has changed, and how substantially some of the software upgrades can alter the database structure, I'm not particularly concerned.
To get to the real question, the earliest non-mediawiki, non-deleted revision is revision 916, made 8 minutes after the mediawiki installation. (The earliest non-mediawiki deleted revision is an unnumbered revision made in the intervening 8 minutes, by WikiSysop at 20050112025710.) The first few dozen edits were all made by Test, after which Daveh finally created a real account (Wrye was given an account a couple days later, followed soon after by various other editors whose names I don't recognize), so from those users' contribution histories you should be able to figure out any other questions. Based on Daveh's history, he probably considers the January edits to be part of the research/testing, in which case he might consider his first edit as Daveh to be the "real" first edit.
The lowest number revision is rev_id 6 -- an edit which is also notable for creating an ugly little problem that took three years to fix.
All of which is probably far more than you really wanted to know. That's what you get for asking questions ;) --NepheleTalk 04:07, 9 August 2010 (UTC)
Hahaha...yes, that's a lot more than I wanted to know, but it was an interesting read nevertheless, especially the "ugly little problem" part. I wondered if deleted pages might play into it, but I couldn't see why Main Page would ever have been deleted. All I had actually wanted was a meaningful date to use as a default for {{ForYMD}}, instead of an arbitrary one, when I stumbled across the strangeness and wondered if it was ominous or benign. Robin Hoodtalk 04:51, 9 August 2010 (UTC)


I don't want to seem pushy or anything, but while I'm looking at your page anyway, would you mind horribly if I archived 2009 for you? This page is getting really really long! :) Robin Hoodtalk 04:55, 9 August 2010 (UTC)

Nope, I don't mind at all. I've been meaning to do it myself but my time has been spent on the server side lately. -- Daveh 02:14, 14 August 2010 (UTC)
Okay, done. I also converted your Archive list to {{Archive Table}} format so it doesn't start running across your screen too much as the years go by. :) Robin Hoodtalk 03:30, 14 August 2010 (UTC)

Synchronization IssuesEdit

I recently noticed a problem with HotnBOThered where I was forced to log in under an old password in order to use it. This had actually been happening intermittently for some time, but because it was an automated process, I didn't understand what was going on. I "re-changed" the password and everything worked fine after that. I've just now experienced the same issue with my main account. Can you or Nephele please look into it? I realize the passwords aren't plain-text, MD5 encoded as I recall, but I suspect the hash should still match amongst all our servers if all is well. Robin Hoodtalk 23:52, 23 August 2010 (UTC)

The passwords should only be stored in the one database we have (db1) so there shouldn't be any issue with old/cached passwords (at least at first glance). By "old" how long ago did you change passwords? Also, for the past month or two we've solely running off of content1 with no squid or other cache mechanism in between servers that might be misbehaving. -- Daveh 00:04, 24 August 2010 (UTC)
With HotnBOThered, it was that way for months before I realized what was going on, and I think I re-changed that password maybe about a month ago. With my main account, I last changed it about a week ago. Robin Hoodtalk 01:14, 24 August 2010 (UTC)


Hi Dave. I thought I'd direct you to this discussion of trademarked images for your opinion since, ultimately, you'd be the one responsible in the end. Thanks! Robin Hoodtalk 19:43, 25 August 2010 (UTC)

Interwiki Link Not WorkingEdit

Hi Dave (or Nephele, whoever gets to this first). Can one of you have a look at why the following interwiki link doesn't work: [[mediawikiwiki:Help:Patrolled edits]]. Looking at maintenance\interwiki.sql, it looks like it should be there in a default MW 1.14 install. Robin Hoodtalk 01:04, 30 September 2010 (UTC)

It was missing from the Interwiki links database. Why it was missing I don't know, but it's there now and appears to work. -- Daveh 02:08, 30 September 2010 (UTC)
Perfect, thanks Dave! Robin Hoodtalk 21:08, 30 September 2010 (UTC)


Daveh, check They present our site with a way I can not describe. It seems that they hate us! They even have a section that lists the most effective ways to vandalize our Wiki! They call us "laughable content"! They insult our most important members including yourself! They have an image "180px-UespLicensePlate.png" in which they make fun out of you! You should take action! --Rigas Papadopoulos Do you want anything? 10:16, 2 October 2010 (UTC)

They're free to post on their site what they want, as long as they credit our work. Everyone is free to express themselves, which allows for some very creative work, but it also means you should accept the fact that less sophisticated people are free to write too. --Timenn-<talk> 11:01, 2 October 2010 (UTC)
We can't do much for this can we? But what do you mean by "as long as they credit our work"? I thought that means that they are helping us not harassing us. Still you are right that they can write at their website whatever they want as long as it is not violating copyright. Which I think they do by using our logos. Anyway, I will wait for Daveh to answer since he is the creator of our Wiki. --Rigas Papadopoulos Do you want anything? 11:09, 2 October 2010 (UTC)
Note that this is exactly what Encyclopedia Dramatica does and is known for. see wikipedia:Encyclopedia_Dramatica. Just check out their entries on some other sites like Wikipedia (Not Safe for Work content, although that probably goes for most of the site). In short, I'd be disappointed if they didn't make fun of us. -- Daveh 12:41, 2 October 2010 (UTC)

So their purpose is to make fun out of everything and because we are a popular site they deciced to target us. That explains everything. I have a question; The image they have is really your car's Licence Plate? --Rigas Papadopoulos Do you want anything? 13:00, 2 October 2010 (UTC)

Yes it is. I posted it originally somewhere here on the site. -- Daveh 13:03, 2 October 2010 (UTC)
Thanks for answering my questions! --Rigas Papadopoulos Do you want anything? 13:07, 2 October 2010 (UTC) P.S. Is the image still on our site?
It was originally posted on the blog. I'm still jealous :) rpeh •TCE 13:33, 2 October 2010 (UTC)

I can not understand why you are "still jealous". What happened? --Rigas Papadopoulos Do you want anything? 14:21, 2 October 2010 (UTC)

He is jealous because he did not get that plate too. It is like a college roommate of mine that had "Styx 1". Brf 14:33, 2 October 2010 (UTC)
Why did rpeh not get it? Any problem with the goverment? --Rigas Papadopoulos Do you need anything? 14:43, 2 October 2010 (UTC)
Reph couldn't get it because Daveh already had it. Brf 14:47, 2 October 2010 (UTC)
You're right. The same thing happens here in Greece and I suppose in every country. Thanks for telling me. --Rigas Papadopoulos Do you need anything? 14:56, 2 October 2010 (UTC)


What does this page contain and is inaccessible by nobody other than you? --Rigas Papadopoulos Do you need anything? 14:45, 2 October 2010 (UTC)

As far as I remember this contains things like the financial records. I could be wrong though. The last time I read anything about this was months ago in one of the archives.--Ghurhak gro-Demril or TAOYes? 17:56, 3 October 2010 (UTC)
See UESPWiki:UespSiteStats. Robin Hoodtalk 20:09, 3 October 2010 (UTC)

I can not understand much but thanks for trying to help! --Rigas Papadopoulos Do you need anything? 13:03, 5 October 2010 (UTC)


Just now I had a problem with UESP, I could not go to any of site's pages while any other site at my computer worked properly. What happened? --Rigas Papadopoulos Do you need anything? 08:49, 3 October 2010 (UTC)

It happens every day at about that time. It's a server restarting or a backup running or something. rpeh •TCE 09:03, 3 October 2010 (UTC)

Thanks for answering me. That never happened to me before so I figured I should ask the creator of the site. Thanks again! --Rigas Papadopoulos Do you need anything? 11:22, 3 October 2010 (UTC)


Heya. I'm a longtime UESP lurker (from before it was a wiki!) and have been contributing to the official and unofficial wikis for Guild Wars. Recent changes at GuildWiki's host are leading us to to consider moving. Do you have any suggestions, advice, things to avoid etc? Thanks much.

AKA Tennessee Ernie Ford 04:36, 4 October 2010 (UTC)

There's a lot of advice I could give you, not all of it would be useful so I'll try to give you a few of the more important points :
  • Your type of server (VPS, dedicated, etc...) depends on what sort of traffic you're getting and what growth you're expecting in the near future. If you have about 500 unique visitors daily (as you mentioned on your site) then assuming this is in the order of 10,000 page requests daily then a VPS should easily handle this sort of load. I don't remember exactly when our VPS provider began to fail but I think it was in the 100k daily page range. A single modest dedicated server can get you up to 1 million daily pages for MediaWiki if properly configured (but this is stretching it). If you are growing rapidly you might consider getting a low end dedicated right away. At the moment we are primarily running off two servers (one content and one database, our squid server is broken) and easily handle 600-800k daily page views.
  • We use iWeb which, while not perfect, has served us very well, both in terms of cost, service, and support. For example, their low end dedicated servers run about 50$/month if you pay for 2 years up front.
  • Server performance depends a lot on setting up Apache/MySQL/PHP/MediaWiki correctly. There is a huge difference in performance between the "wrong" and "right" settings. We've learned through experience (i.e., using the wrong settings at first and figuring it out through trial and error).
  • Google AdSense has served us very well in terms of covering all operating costs. Currently it is generating more than twice our costs which gives us a good amount of money left over for emergencies or future growth. And this is just with the one small ad at the bottom of the page which is typically one of the average performing ad locations (compared to an ad at the top of the page).
  • Monitoring your site's performance is key in determining whether you have room to grow or if you're right at the edge of what your server can do. It is also helpful for those random server outages.
  • Backups are also critical (for example, we've had two hard drives fail over the years, one on our database server). Make sure to keep off site backups as well (if you just backup locally and your hard drive fails then both your data and backups are gone) and check/test your backups to make sure they are working.
At any rate, good luck. Wikia offered us hosting several times in the past and what they're doing is the exact reason we declined. -- Daveh 01:54, 7 October 2010 (UTC)
Thanks for the reply (I seem to have missed it when it, so apologies also for the delayed ty.) 15:48, 21 October 2010 (UTC)


Many times, when I look at talk pages I see grammar mistakes and I fix them. However, if do this I get a message, (usually from S'drassa that tells me it is not allowed to edit other posts by other people. Is there any way I can get the right to do that? I ask this because 1) I do not change the meaning of the post 2) Because I link the post with pages that the person that wrote it did not do it 3) Because I fix spelling which gives the page a better look. The sentence "what is the cheapest destruction spell that is used at novice level. i need to use it over and over again to get my skill level up." looks much better if it is written like this "What is the cheapest Destruction spell that can be used at Novice level? I need to use it over and over again to get my skill to level up." The example was taken from the Oblivion Talk:Destruction page. --Rigas Papadopoulos Do you need anything? 13:18, 5 October 2010 (UTC)

A Talk page is just a Talk page, like a chat. You would not expect to correct everyones spelling and grammar in a chat would you? --Brf 13:21, 5 October 2010 (UTC)
I like doing that. That's why I am asking permission to do it. I have no problem to look at Talk Pages all day to find and fix wrong spelling. --Rigas Papadopoulos Do you need anything? 13:38, 5 October 2010 (UTC)
You are missing the point. The patrollers do not correct other users' posts on Talk pages. --Brf 13:47, 5 October 2010 (UTC)
I did not say I want to become a Patroller. I just want to do what I mentioned above. --Rigas Papadopoulos Do you need anything? 13:50, 5 October 2010 (UTC)
You are still missing the point. A talk page is just that. If I misspell a word or use bad grammar, it causes no harm. There is no need to correct these, since a talk page is not the same as an article. I am not presenting information for future readers. I am simply communicating a point or asking a question. By appending my signature, I am showing that these are my own words, not the words of this site. If someone came along and editted those words, they would be the words of the editor, and not mine --Brf 13:57, 5 October 2010 (UTC)

() This discussion ends here. Rigas, you are NOT allowed to edit the posts of other users. I am an administrator, and I am not allowed to edit or correct the grammar of other users. All posts from now on will be moved to your own talk page. --Krusty 14:00, 5 October 2010 (UTC)

Cartographer RightsEdit

Can I get Cartographer rights, please, so that I can fix issues like these when they pop up? Thanks! Robin Hoodtalk 23:26, 7 October 2010 (UTC)

I'd prefer that Cartographer rights were kept to people who helped with the map. We had problems with one other editor who was given the rights and didn't understand what he was doing. Of course, it's up to you. rpeh •TCE 08:11, 8 October 2010 (UTC)
I'm fine either way, but I figured that with GK and Timenn being only partially active lately, it might be a good idea to have more than just rpeh as the only regularly-active Cartographer. On the other hand, it's not like the map data is terribly time-sensitive. Robin Hoodtalk 08:27, 8 October 2010 (UTC)
Actually, I'm probably overreacting. RobinHood70 has his head screwed on and it's not as if there are going to be dozens of changes to make. If you're adding people, it'd probably be a good idea to add Krusty too, since he's an admin. I'm sure both of them will be capable of answering the question "Did you change anything else?" accurately, which was the problem I mentioned earlier... rpeh •TCE 17:18, 8 October 2010 (UTC)
I think it's OK to have them both added to the cartographers, seeming as I trust both know that changes to the map aren't automatically reversible. As for the availability... I'm active enough to notice change requests for the map, if rpeh doesn't beat me to it. --Timenn-<talk> 09:00, 13 October 2010 (UTC)
Done -- Daveh 19:17, 17 October 2010 (UTC)
Thank you! Robin Hoodtalk 03:07, 18 October 2010 (UTC)


I hope you don't mind, I will create a page at Wikipedia about the site. --Rigas Papadopoulos • Talk to meMy work 13:35, 11 October 2010 (UTC)

I'm pretty sure there was one a while ago and it failed to meet notability requirements or something, so it was deleted. I don't know if that would still be the case—I'd have to read over their guidelines again—but I thought I'd mention it in case someone puts it up for deletion. Robin Hoodtalk 20:57, 11 October 2010 (UTC)
It's not that we would mind (we don't) but whether Wikipedia minds since it is their site and content. While we run the same software (MediaWiki) there is no relationship between the two wikis. -- Daveh 00:47, 12 October 2010 (UTC)
I know that the previous article was deleted, so I am delevoping the new at user space and I will post it after I finish. I also hope you do not mind about this one, I am going to post information abut some of our most important members. --Rigas Papadopoulos • Talk to meMy work 07:36, 12 October 2010 (UTC)


Could you please change access from Admins and Patrollers to Admins, Patrollers and Userspace Patrollers? There's an edit in userspace I can't patrol because it's the first edit and there were two more after it.--Ghurhak gro-Demril or TAOYes? 19:48, 12 October 2010 (UTC)

Can't you access the link "Mark as Patrolled" that is present in the lower right corner? It should be there when you open the first revision (the N) from the Recent Changes. --Timenn-<talk> 08:53, 13 October 2010 (UTC)


I just did an edit at a fully protected page! The page is UESPWiki:Site Support/September 2010. Is this normal? As far as I can remember I am not an Administrator. What happened? --Rigas Papadopoulos • Talk to meMy work 18:53, 17 October 2010 (UTC)

Evidently...I didn't setup the page protection templates and can't exactly see how they work, although the site protection is missing a category link although that appears to be on purpose. -- Daveh 19:11, 17 October 2010 (UTC)
I can not understand. Can you use less complicated English please? Remember, I am a 12 year old boy from Greece. --Rigas Papadopoulos • Talk to meMy work 19:15, 17 October 2010 (UTC)
You did nothing wrong there just appears to be an issue with how we protect those pages. I or one of the other admins will look into it. Thanks. -- Daveh 19:17, 17 October 2010 (UTC)
I now understand. I hope I proved useful at detecting the problem. :) --Rigas Papadopoulos • Talk to meMy work 19:22, 17 October 2010 (UTC)
I'm on it. --Krusty 19:52, 17 October 2010 (UTC)

New namespaceEdit

Can you create a "Stormhold" namespace? Every other game has. --Rigas Papadopoulos • TalkDeeds 10:40, 24 October 2010 (UTC)

You Might Want to Take a Look at...Edit

The blog. It's got a ton of comments backed up and you're the only one that can approve/reject them.--Ghurhak gro-Demril or TAOYes? 23:09, 26 October 2010 (UTC)

Yes, thanks for the reminder. -- Daveh 00:32, 28 October 2010 (UTC)
You're welcome!--Ghurhak gro-Demril or TAOYes? 00:35, 28 October 2010 (UTC)

Recent Changes ProblemsEdit

We're getting some odd behaviour from the wiki. It seems possible it's from Content1, but that's not absolutely certain. See this message. Robin Hoodtalk 06:04, 31 October 2010 (UTC)


Something is wrong with this page. Half of the times I visit it while logged-out, at the bottom of the page, right next to the message that says when the page was last modified, a new message shows up that says how many times the page has been visited. That never happened to me with any other page. This happens only when I am logged out. What's happening? --Rigas Papadopoulos • TalkDeeds 16:57, 31 October 2010 (UTC)

It may have just been due to the recent return to using Squid caching. Check the page again and see if you still have the same issues. -- Daveh 19:46, 31 October 2010 (UTC)
Nope, I tried four times but I can't get it to happen again. So I suppose that the "Squid caching" is responsible (I have no idea what the Squid caching is). --Rigas Papadopoulos • TalkDeeds 19:53, 31 October 2010 (UTC)
I changed a setting earlier this afternoon that should have fixed the issues you and a few other people were seeing. You don't need to worry about what "Squid caching" is but the simple explanation is that it will reduce our server load and let us "do more with less". -- Daveh 21:01, 31 October 2010 (UTC)


I downloaded this file and uploaded it to Wikipedia. Are there any copyright issues? --Rigas Papadopoulos • TalkDeeds 14:53, 1 November 2010 (UTC)

Shouldn't be unless any of the other admins/editors can think of something. -- Daveh 17:49, 1 November 2010 (UTC)
It currently seems fine, check this. If there is anything missing or wrong, add it please. --Rigas Papadopoulos • TalkDeeds 17:54, 1 November 2010 (UTC)

Minor Site Config IssueEdit

I noticed today while testing something that when you're logged in and type into your browser, the address in your browser bar is automatically redirected to display Okay, so far so good. But then if you log out and do the same thing, the main page still appears, but your browser's address bar will still just say I don't know if anything can or should be done about it, but since it differs from how works, I thought I'd mention it. Robin Hoodtalk 02:19, 3 November 2010 (UTC)

I would like to congratulate you for making the least useful post EVER on UESP. *Applause* 02:58, 3 November 2010 (UTC)
Why is this an issue and what do you expect Daveh to do about it? rpeh •TCE 03:03, 3 November 2010 (UTC)
Since I'm editing anyway to remove the second comment (see history), I figured I'd ask as well; is there any reason to expect that this will cause any problems or that "fixing" it would make any difference? --GKtalk2me 03:08, 3 November 2010 (UTC)
No, so far as I'm aware, it doesn't cause any significant issues, other than the fact that DotNetWikiBot (and possibly others) try to identify the "/wiki/" short-form and won't find it properly without the redirect. Even then, it causes no major issues, though. I only mention it because of that minor point, and the fact that it may indicate a configuration issue that would have other effects I'm not aware of (e.g., unexpected behaviour for IP editors). Robin Hoodtalk 04:16, 3 November 2010 (UTC)
While it appears to be a minor issue it's good to know about these things. In some cases it may be a misconfiguration that may cause larger issues down the line. -- Daveh 12:20, 3 November 2010 (UTC)

Serious Login ProblemEdit

Hi, I am this guy. The problem is, I can't login. If I try to login, no error message or something appears, but I stay logged off. This may be a serious problem. Can you check it? Thanks. -- 06:07, 4 November 2010 (UTC)

I'm TheAlbinoOrc. Same thing! What's going on?-- 06:09, 4 November 2010 (UTC)

The iWeb data center that houses all of UESP's servers had a big power outage last night due to a nearby fire. All servers but content2 are back up and I just fixed the NFS disk mounting issue on content1 which was causing the login issue. I'm currently traveling in Puerto Rico and have limited time/means to look into the issue in more detail at the moment. -- Daveh 12:14, 4 November 2010 (UTC)
Thanks Dave. I'd noticed content2 was down and assumed you'd been doing some upgrades. Things seem to be working now - thanks for the fix. rpeh •TCE 12:27, 4 November 2010 (UTC)

Another Recent Changes problem...Edit

Whenever I try to go the Recent Changes by typing the URL "" while logged out, I don't end up on the current Recent Changes, but at an earlier version, the version that was there at the time when I last visited the Recent Changes by typing the URL. Just now, I ended up at a version of November 2, and the most recent edit was my response to Bomb Bloke's comment at the Oblivion:Glitches/Proposed#Teleportation. That happened many times, and only happens while I am logged out. Any ideas? --Rigas Papadopoulos • TalkDeeds 21:30, 4 November 2010 (UTC)

If you view pages when logged out you get a cached copy of a page, which could be an out-of-date one. To make sure the page is the latest version, just log in. rpeh •TCE 08:04, 5 November 2010 (UTC)
Thanks. --Rigas Papadopoulos • TalkDeeds 08:16, 5 November 2010 (UTC)

Is that you?Edit that you?

Mor'tar'iit 12:19, 7 November 2010 (UTC)

Nope! God doesn't need wikis--Ghurhak gro-Demril or TAOYes? 23:42, 12 November 2010 (UTC)

Map Caching IssueEdit

The latest reply in the thread about watering the MW map is worrying. Is it possible that the caching change you made is causing this? I must confess the weirdness that I've been seeing on the test map with a mixture of OB/MW places appearing has had me worried - I don't think I've screwed up the test map but my PHP-fu is weak.

TL:DR - please take a look at the test map and let me know if this is caching or an rpeh-screwup. rpeh •TCE 23:28, 12 November 2010 (UTC)

We've been having minor DNS issues over the past day or two, or, more accurately, has been having problems. This has caused the sub-domains to randomly not get resolved. Right now, for example, content1 and maps don't properly resolve which is why the map images don't display. If you wish to temporarily use IP addresses for testing files1/maps is and content1 is (although using IP addresses might not work due to using virtual domains for the web servers on files1). -- Daveh 00:21, 13 November 2010 (UTC)

Cite ExtensionEdit

If you have a few spare minutes, please could you install the latest version of the Cite extension? (Well... the MW 1.15 version, anyway). It's not critical, but the latest versions allow a much cleaner way of defining references (see the "Separating references from text" section on that link), which would make a number of pages much easier to edit - Lore:Septim Dynasty for instance. There's absolutely no hurry, but it would be handy to have. rpeh •TCE 15:48, 13 November 2010 (UTC)

No problem...I'll look into it. -- Daveh 16:14, 13 November 2010 (UTC)
It's still not critical, but just in case you've forgotten... nudge. rpeh •TCE 01:45, 17 January 2011 (UTC)
Bah, sorry. I'll look into it 'shortly'. -- Daveh 00:45, 18 January 2011 (UTC)
Thanks. And don't call me 'shortly'. rpeh •TCE 07:29, 18 January 2011 (UTC)
Installing the v1.15 of Cite seemingly doesn't work with the v1.14 of MediaWiki we're running and I've been unable to download the v1.14 of Cite due to some subversion repository error there. I'll try again in a few days although the v1.14 may not have the feature you're talking about. -- Daveh 21:46, 22 January 2011 (UTC)
Damn, I forgot we were still on 1.14. For some reason I thought we were on .15. Please try the latest 1.14 version if you can, and I'll hope it's got the new feature in. Otherwise it'll just have to wait until the next MW upgrade. rpeh •TCE 08:18, 24 January 2011 (UTC)
That feature wasn't added until revision 53790, so you'd need the 1.16 snapshot. Robin Hoodtalk 09:19, 24 January 2011 (UTC)
I'll try the v1.16 and see if it cooperates. At some point we should upgrade the entire Wiki but in the near future there are a few other major things to work on first. -- Daveh 14:14, 25 January 2011 (UTC)

Site PerformanceEdit

The site has been going really, really slowly for the past 4 hours or so. I can't see anything on the stats pages that would point to anything. Can you spot anything on the servers? rpeh •TCE 12:51, 16 November 2010 (UTC)

I noticed that too earlier this morning although it seems fine now. It wasn't squid1 as both content1/2 were slow which seems to point towards db1 although there were no obvious issues there either. Looking at the network logs I think the issue might have been re-enabling the live database replication on content3 last night (I haven't updated the upgrade history page yet regarding this). According to the network history it took all night to catch up and stopped earlier this morning. This would have put a higher load on db1 which is the likely candidate. Now that the content3 db replication is caught up there shouldn't be further issues. -- Daveh 14:26, 16 November 2010 (UTC)
Definitely all fine again now. Thanks for checking. rpeh •TCE 14:32, 16 November 2010 (UTC)

Another ProblemEdit

Sorry to bother you again but we're getting what looks like a caching problem: some diffs are only showing on one server. Compare a diff on content1 to the same one on content2 (or content3). Or another one.

It's an annoyance rather than a problem, since the changes are applied everywhere, but obviously a fix would be nice. rpeh •TCE 09:25, 17 November 2010 (UTC)

More information on this here. The old image upload problem has come back, too - images when uploaded will sometimes only be visible on one server but not the other. A purge on the problematic server fixes the problem, though. rpeh •TCE 00:27, 21 November 2010 (UTC)
The diff issue was due to using separate instances of memcached on content1/2. I've set it so there is only one memcached instance on content1 which content2/3 will use. While this makes content2/3 a bit slower but we're not anywhere close to maxing out the content servers and it should fix two issues you mentioned. I haven't tested the image issue but I assume it was due to the memcached settings as well. Note that content3 isn't completely setup yet and currently has some issues serving pages (due to firewall settings on content1 I think). -- Daveh 14:38, 21 November 2010 (UTC)
That looks to have fixed both problems. Thank you. rpeh •TCE 16:55, 21 November 2010 (UTC)


(moved to User_talk:Rigas_Papadopoulos#Namespaces)


After a recent conversation in IRC I discovered that the UESPMail server is no longer working. This is the URL I had, but now it delivers a 404 error. I don't know it this is deliberate or an accident, but I thought I should mention it. rpeh •TCE 02:18, 29 November 2010 (UTC)

Thanks...this is due to the restore of content3 which was hosting the email. I'm still trying to fix a stubborn PHP session error there but getting email working is the next thing on my list. -- Daveh 20:12, 30 November 2010 (UTC)
Hey Dave - I have not been able to log onto my UESP mail for quite some time now. I have no idea if it is related to rpeh's request from november, but maybe you could look into it? In the meantime, I'll shut down my mail-link. Thanks! --Krusty 18:23, 8 January 2011 (UTC)
Mail is unfortunately resisting my attempts to install and restore the configuration from backups. Combine that with lack of time on my part from holidays and other projects means it is taking a lot longer that I'd like. Hopefully this isn't causing much issue to the few using the mail service. -- Daveh 16:31, 11 January 2011 (UTC)

Morrowind MapEdit

I think the new tiles for the Morrowind map are ready to be moved onto the live server from the test box. Before you do it, please could you cast your eye over this topic and look at the outstanding issues - the odd lines and strange water. Personally I think the improvement of the colourised water is far more important than a few oddities but I'd like your opinion. I'd also like to hear what you think about the sharpened zoom17 tiles. If all is okay, it's ready to go! Thanks. rpeh •TCE 13:18, 5 December 2010 (UTC)

While you're online... nudge! rpeh •TCE 21:24, 12 December 2010 (UTC)
Thanks...I had missed/forgotten this. -- Daveh 21:27, 12 December 2010 (UTC)
Done, although appears to be having load issues not related to this change (it is rather slow and sometimes timing out). -- Daveh 21:44, 12 December 2010 (UTC)
Thank you! One extra thing: please can you copy images/mwblank.jpg and images/mwoutofrange.jpg over too? They make it look a little cleaner. Otherwise, we now have water, water everywhere!
I had been wondering whether or not to post about the speed issue. I've been getting delays on for the past couple of hours, and that's obviously the same server as files1. Is there a problem here? rpeh •TCE 21:57, 12 December 2010 (UTC)
I think I fixed the files1 performance issue. Just a misconfigured lighttpd which was hitting the max connections due to the increased traffic over the weekend (around twice the normal traffic). -- Daveh 23:19, 12 December 2010 (UTC)
It's certainly looking better now. Thanks! rpeh •TCE 23:34, 12 December 2010 (UTC)

The elder scrolls V skyrimEdit

i have some extra additional info i think this is the real magazine and i found it online

Prev: Archive 2009 Up: User talk:Daveh Next: Archive 2011
Return to the user page of "Daveh/Archive 2010".