GuildWiki talk:Community portal/Archive 17

GuildWiki... Disappointment surfing
I'll be short on this post... many people must know about this issue, GuildWiki servers need to be updated to be capable of maintaining all the ammount of people that surf its web pages.

If funding is the problem, they can at least set up a paypal account for us to make easy donations and they might get enough funding to buy a descent connection.


 * You could also at least read the entire page, and you'd know that their problem is more they refuse to accept and utilize help and resources that would benefit the wiki. ErkDog 06:59, 15 November 2006 (CST)


 * I am glad now to surf on wiki's pages as they finally added a new server =]. Solito 14:28, 16 November 2006 (CST)

10000 Articles!
I just noticed that we've broken through the 10000 article barrier. I'd say a prominent note on the main page and a big Thank You to our contributors is in order for reaching this milestone?! -- 10:18, 25 September 2006 (CDT)
 * I've drafted a message at Main_Page/site_notice. I've also mocked up Main Page/editcopy to show how it would look.  If that's what you had in mind, let me know and I can re-activate the site notice banner on the Main Page.  --- Barek (talk • contribs) - 11:11, 25 September 2006 (CDT)
 * That looks very good. And ... err ... yay us! --Xeeron 11:30, 25 September 2006 (CDT)
 * I went ahead and added the notice to the Main Page. --- Barek (talk • contribs) - 11:37, 25 September 2006 (CDT)
 * Nice! (Unfortunately, there are at least 400 pages of junk in the wiki...) ~ Nilles (chat) 12:18, 25 September 2006 (CDT)
 * Does it count user name space stuff? Or only main name space? --[[Image:Gem-icon-sm.png]] (talk) 13:21, 25 September 2006 (CDT)
 * I was simply referring to Untested builds and Unfavored builds - about 80% of then or so are junk that might work in game but actually are not meant for documentation. In fact, I'd simply delete and abandon these pages for the sake of simplicity. After all, new players who search for new builds are in search of new ideas and a grasp of uniqueness. We don't serve them by displaying builds like Me/Mo Faster Caster - builds, that any of those players could easily invent themselves. ~ Nilles (chat) 16:01, 25 September 2006 (CDT)
 * I would still like to know the answer to the question. :) --[[Image:Gem-icon-sm.png]] (talk) 16:29, 25 September 2006 (CDT)
 * GuildWiki's Special:Statistics page doesn't specifically mention user pages. If that query is universally consistent across all MediaWiki installations, then user pages are not included per Wikipedia's Special:Statistics.
 * To be certain, we could always test it and delete the test-created user page afterwards (being certain to refresh the cache to ensure the numbers shown are correct). --- Barek (talk • contribs) - 16:33, 25 September 2006 (CDT)


 * As far as I'm aware it only counts articles in the main name space that are not classed as short articles.  &lt;LordBiro&gt;/&lt;Talk&gt; 19:09, 25 September 2006 (CDT)
 * Hooray for the wiki! This is a major milestone in its history - now let's shoot for 20 thousand, eh? ;) [[image:ordinsig.gif]] Ordin 20:45, 28 September 2006 (CDT)
 * Based on the rate of article growth following the release of Factions, I suspect that we'll be well over 12,000 pages by November sometime :-)  --- Barek (talk • contribs) - 20:49, 28 September 2006 (CDT)


 * As Biro said, it's articles in the main namespace. But they also have to contain a link to another article, so dead-end articles don't count.  I think it might also have to meet a minimum length, but I'm unsure and too lazy to look at the code right now.  --Fyren 20:50, 28 September 2006 (CDT)

State of the Wiki: Builds
I argue that as an encyclopedic reference to Guild Wars not including popular builds would be very narrow-sighted. Now how this would be done is pretty tricky. Shandy 07:11, 18 November 2005 (UTC)

That was one of the first answers I got when I started talking about builds on the wiki. And Shandy was right, including builds has worked out, but it is tricky indeed. So for all those who are not involved in the builds process (but often see lots of build related pages spamming recent changes), let me give you a quick overview what the build section currently does, why there was a lot of talk about it in the last months, what is not running smooth and why adding builds to the wiki is still a great thing.

History
The wiki was founded by a crowd of PvE lovers, so after the first months, there was decent coverage of most PvE aspects of the game, but the builds section was almost entirely consisting of the pre-builds, plus several weird concept builds. To sort out those of them worth keeping, a build vetting process was introduces. Later, when people started to post more new builds as well, that evolved into the voting, which is more or less in the same form still in use.

The new section drew some attention from those who possess superior (to mine) wiki skills, so soon we got nice templates, ordered categories and a beautiful builds portal. Unfortunatly, as the number of people working on the wiki to test builds grew, the number of new builds exploded. That lead to several attempts to streamline, change or speed up the process of build vetting, most of which did not have a big impact.

Right now, a refined version of the basic voting vetting is still in place, I tried to write that (and all other relevant stuff about builds) down here.

The basic problem with builds
The basic problem with builds (and reason behind 95% of all troubles related to builds) is the nature of builds as subjective articles. Unless almost all other aspects of the wiki, the quality of a build can not be objectively measured. That means, if there is to be any skale among builds, it will be a subjective one. That of course can give rise to many conflicts: Either if people disagree about builds, and, maybe even harder to solve, there is a conflict about the correct mechanism of how to take lots of subjective opinions and synthesize them into one wiki article. The later lead to long discussions about a build policy, without any consensus reached yet, meaning the wiki currently does not have any build policy.

Why are builds on the wiki still a great idea
Despite the frustration which can arise about all the conflicts related to builds, this should not distract from one fact: The users love them! The build portal currently gets around 15.000 views per day, making it the 5th most popular page of the wiki. The guildwiki has proven that builds do not only belong on forums, but that the wiki concept can be used to provide simple and easy access to build ideas to all users. The builds section has also helped to attract many new users to the wiki, especially from the previously hardly represented PvP part of the community.

Let me put in 2 maybe somewhat specific lessons that can drawn from the builds part of guildwiki:
 * Templates and nice structure works, even for new users, as long as the user side is kept simple. There is possibly no template in the wiki which is more successful than Template:Skill bar. I have seen literally hundreds of new build articles, done by first time users, and more than 99% of them understood and correctly used the template (and the later attribute template as well) for their first build. As long as there is a page describing their use and the template keeps using plain text on the part the user sees, they do work, making articles much nicer to read.
 * Users like portal pages. The builds portal consists to 80% of just a listing of Category:Builds, yet after it was introduced, the number of pageviews skyrocketed. New users unfamiliar with the workings of a wiki will have a hard time if they hit a link on the main page and are dumped in a category with a one line description. The best information is not good enough, if users cant find it. Some other parts have great portal articles as well (for example the green weapons lists), but there are still several links on the main page which lead to ugly categories. Some beautification of these links would maybe be one of the simplest ways to enhance user value in the wiki.

So in the end, builds on the wiki work, not without some problems and effortless, but the result proves that the effort is well invested. --Xeeron 10:46, 6 October 2006 (CDT)

Discussion
It would be such a shame to break up the above with comments, so I'll put mine here instead :) This is a good summary for anyone unfamiliar with the Builds section of the wiki who is interested in learning about the recent discussions. Good work Xeeron! You've kept it pretty objective, so I don't have any problems with it.

It is certainly true that portals in general are generally well received, and for some time I've been an advocate of creating articles to compliment the current category setup that we have on the wiki.

Anyway, good work Xeeron! <- for you.  &lt;LordBiro&gt;/&lt;Talk&gt; 11:34, 6 October 2006 (CDT)


 * I am continually amazed by you Builds people. :) &mdash;&mdash;Tanaric 02:17, 7 October 2006 (CDT)

Halloween skin
I have an idea, maybe we could do the site up in orange and black for halloween? :)

I'm thinking of a crossed flamberge over pumpkins for the logo, and replacing the white with black and blue with orange for the week leading up to halloween. Mad king background instead of books? :p Any thoughts? &mdash; Skuld 10:09, 9 October 2006 (CDT)


 * I am willing to do some work :) I'll make some drafts and throw down a link here :) -- Ifer (t/c) 10:27, 9 October 2006 (CDT)


 * Sounds fun. Has there been any news about a Halloween event this year? I think it was announced last year about this time, iirc. &mdash; Gares 10:30, 9 October 2006 (CDT)


 * Yeh, i'll try find the gaile log brb &mdash; Skuld 10:33, 9 October 2006 (CDT)


 * http://www.guildwarsguru.com/forum/showthread.php?t=10056831 &mdash; Skuld 10:51, 9 October 2006 (CDT)


 * Thx. Can't see GWG from work, but I'll read about it when I get home. &mdash; Gares 11:01, 9 October 2006 (CDT)


 * If you make it the default skin/CSS, I will kill everyone involved. --Fyren 16:50, 9 October 2006 (CDT)
 * And I'll dismember and burn the remains. This sort of thing should be an optional skin - not changing the default ones. --- Barek (talk • contribs) - 17:07, 9 October 2006 (CDT)

before we let your edit thru
As a protection against automated spam, you'll need to solve this equation before we let your edit thru: 

Could someone PLEASE take care of that horrible english? -- Ifer (t/c) 10:34, 9 October 2006 (CDT)


 * I changed thru to through, do you want to message altering in any way? &mdash; Skuld 10:51, 9 October 2006 (CDT)


 * Thanks a lot :) That has annoyed me since I started working on the wiki ;) -- Ifer (t/c) 11:05, 9 October 2006 (CDT)

Skill box revamp
I've proposed a bunch of changes to the skill box template (and been posting about it for about ten days at the talk page linked above). This includes parameter name changes, auto-generation of progression tables, slight display tweaks, and factoring out all the CSS into MediaWiki:Common.css. Each topic has its own section at the linked talk page. I plan on making these changes soonish so they're live before Nightfall and allowing for at least a couple days for changes/fixes. Please comment at there. --Fyren 06:04, 15 October 2006 (CDT)
 * Functionality-wise I don't have any objections. It looks like you've planned well and the use of the new template is easy enough to grasp with examples at hand. The only thing left to do before implementation is probably writing a style sheet to adjust the looks of it, right? ~ Nilles (chat) 07:18, 15 October 2006 (CDT)
 * I did, but it's not site-wide (yet). The second paragraph in the CSS section describes how to apply it for yourself.  --Fyren 07:40, 15 October 2006 (CDT)
 * As just an ordinary user, not a contributer, I myself find the current changes to the skill summary pages detrimental compared to the version that existed at the beginning of October. I rely on the skill summary pages to allow me to quickly see the textual descriptions of all the skills for a specific attribute or class as well as the cost/casting/reset information.  I recognize it's a work in progress at the moment, but can the finished version please include the textual description of the skill's effect? --IzzionSona
 * Not sure what you're saying. The same descriptions are still there and in the same place.  The same stats are still there and in the same place.  The progression tables were moved up, but they still have the same data.  Can you be more specific?  --Fyren 00:28, 20 October 2006 (CDT)
 * If you meant the quick reference pages, I've fixed a (somewhat major) copy and paste error. --Fyren 01:03, 20 October 2006 (CDT)
 * My comment was the quick reference pages, thank you for the modification to them. And much thanks to all of you who put in the time and effort to make this wiki successful.  It's a huge boon to my Guild Wars experience --IzzionSona

Ok, so I'm new.
I'm new here and I'm a little bit confused on how to post a build. Could someone fill me in, please?
 * Check the links near the bottom of builds. --Fyren 01:08, 20 October 2006 (CDT)
 * those links should help enough.. If it does not, you can ask me. One more thing: sign your comments on talk pages with four tildes, ~, and it will result in this : Ifer (t/c) 03:11, 20 October 2006 (CDT)

Brainstorm: Give Builds their own namespace
There is a difference between Builds and all other types of GuildWiki articles on a fundamental level. The difference is sufficient that I think it can warrent its own namespace (PvE builds and PvP builds are all sharing the same namespace, so I'm not trying to do any sort of PvP segregation here). For example (picking the first build name I see from Recentchanges), "R/Rt Brutal Needler" would become "Build:R/Rt Brutal Needler". This would help the filtering of Recentchanges, Watchlist, and built-in search functions which categories cannot emulate. Of course, simply naming articles that way won't create the namespace (for the purposes of recent changes etc), we'll need Gravewit to fiddle with the mediawiki settings to get the namespace to work. I want to toss the idea out and see what ppl in general think before presenting it to Gravewit. So, care to comment? - 12:12, 22 October 2006 (CDT)
 * I like it - we should've done this from the very start. --- Barek (talk • contribs) - 13:08, 22 October 2006 (CDT)
 * Great idea! Iäm all for it! \o/ --[[Image:Gem-icon-sm.png]] (talk) 13:10, 22 October 2006 (CDT)


 * That's a very cool idea. ~ Nilles (chat) 13:45, 22 October 2006 (CDT)


 * This is exactly the situation that namespaces were designed for. We can discuss how we do this on our side further, but I'll make the technical request to Gravewit now, so we have the structure in place when we're ready. &mdash;Tanaric 15:07, 22 October 2006 (CDT)


 * I am sure I have seen this idea voiced before, no idea why it didnt get taken up back then. --Xeeron 17:10, 22 October 2006 (CDT)


 * Tetris suggested it. --Fyren 19:11, 22 October 2006 (CDT)


 * This will probably mean more incorrectly named builds :P That's ok there's always plenty of us happy to help out with that and the benefits definately outweight the small negative, I'm all for it. --Xasxas256 20:59, 22 October 2006 (CDT)


 * I'm not opposed to this idea but I'm fairly certain it was discussed previously. I wasn't very active at the time when builds were first introduced, so I'm not 100% sure where the information would be, but before anything is implemented I think it would be a good idea to find out why this wasn't implemented initially. Anyone have any suggestions as to where to look?  &lt;LordBiro&gt;/&lt;Talk&gt; 04:39, 23 October 2006 (CDT)


 * I know what you're talking about, I think I recall something along the lines that it was too much work. I think we just failed to find concensus and it got lost in the discussion. Like "usual". ~ Nilles (chat) 05:55, 23 October 2006 (CDT)


 * Actually, Honorable Sarah seems to have first raised the idea: It is both here. Cant really see any reasons for not implementing it there, seems like all people who cared enough to do it didnt follow the discussion. --06:04, 23 October 2006 (CDT)


 * Both Build: and Build talk: are up and running. Can we get a bot working on this? Build should stay where it is, but Builds should be moved to something like Build:Main Page. The redirect should stay more or less permanently, as other sites link directly to the builds portal. &mdash;Tanaric 09:55, 23 October 2006 (CDT)


 * I've moved Builds to Build:Main Page (note, the namespace is sigular form of Build). That seemed a better name than something like Build:Portal, but others can discuss.  Also confirmed that the new namespace is working.  --- Barek (talk • contribs) - 10:09, 23 October 2006 (CDT)

New "Build:" namespace - what are the next steps?
We have nearly 400 builds just in untested, plus all of the tested builds to migrate. I supposed the first question is, does anyone have a bot that could be used to move all builds to the Build namespace? After that, all the auto-generated redirects - do we keep them, or purge? If no one has a bot available for this task, anyone have a suggested method to break this down into workable sized pieces? It could go fast if a group hit it all at once. Last, but not least, we also need to review all of the build policies and guidelines to ensure they begin directing people to create builds with the "Build:" prefix so that they all land in the appropriate namespace. So ... how do we start this process, and what are our next steps? --- Barek (talk • contribs) - 09:55, 24 October 2006 (CDT)
 * Fryen keeps a bot around - 15:53, 24 October 2006 (CDT)


 * On the subject of bots (although pretty much unrelated to the topic at hand) I've wanted to make a bot for a while now. Not because I would do anything different to what Fyren does with his bot, but simply because then I could name it "LordBiRobot".  &lt;LordBiro&gt;/&lt;Talk&gt; 17:56, 24 October 2006 (CDT)
 * I'm going to create a shoepuppet account just to squat on that name!!!!! d-: - 18:56, 24 October 2006 (CDT)


 * Why do you hate love? :(  &lt;LordBiro&gt;/&lt;Talk&gt; 05:16, 25 October 2006 (CDT)
 * What makes you think I hate love instead of loving hate? d-: - 11:36, 25 October 2006 (CDT)
 * Well given that there are 500+ pages involved, this very much calls for a bot. Btw, it would be good to keep the old names around as redirects, unless someone is willing to update the links manually on all of them (or this can be done by a bot as well). --Xeeron 04:45, 25 October 2006 (CDT)

I can do it, but not immediately. After doing the skill stuff, it's pretty clear the Perl framework I use is gimpy, leaving only pywikipediabot as something that'll work for heavy lifting. --Fyren 11:55, 25 October 2006 (CDT)
 * * bump * ... okay, I know the server has been top priority. I just wanted to re-flag this so it's not forgotten, and can get re-added into the to-do list now that the server appears to be responsive again. --- Barek (talk • contribs) - 15:17, 16 November 2006 (CST)
 * I'll probably move all the pages tonight and categorize the redirects it creates into Category:Build redirects to make them easy to find. I'll probably have it move anything in our tested/untested/stub categories, so it'll miss anything not in those.  I'll look into fixing existing links/deleting the redirects later.  If at some point we delete the redirects, it may break bookmarks for people.  --Fyren 19:33, 16 November 2006 (CST)
 * Moved everything, changed category sort keys for categories in builds to drop out the namespace name. However, the categories added by templates like untested-build don't do this, so they should be edited to do so (or all builds will show up under "B").  Documentation should be updated and any new builds made in the main namespace should be manually moved and the redirects flagged for deletion, not added into the redirects cat.  --Fyren 05:02, 17 November 2006 (CST)

Guild Wars Nightfall Bonus Music Key
How do I enter my bonus music key? It's only 15 slots long, whereas every other key is 25 slots long. How do I enter the key so it registers?
 * I don't have mine yet (I know it was shipped out yesterday, hopefully I'll have it tomorrow), but if it is like the other music keys, you don't enter it in the actual game. There should be a web address to go to (www.directsong.com/something), and you'll enter the code there.  You'll then have to download a file (or multiple files) and run them to get the music.  See the DirectSong article for more info.  --Rainith 19:54, 26 October 2006 (CDT)

Another milestone: 10.000.000 visits!
It must have happend about yesterday, but nevertheless: Main Page recieved its 10.000.000th page view. Grats, everyone. --MRA 15:51, 26 October 2006 (CDT)
 * And if you look at the List of Largest Wikis (actually largest MediaWikis), you'll see that GuildWiki has three times as many total pageviews as the English Wikipedia. How did that happen??  We're also 112th overall when measured by "good" page count, 29th by users, and 16th by images.
 * Something has always struck me as off on the page view tally - there's no way we're beyond the english Wikipedia on that. At least one of the sites (theirs or ours) must be counting those differently - or maybe Wikipedia resets theirs periodically.
 * On "good" pages - if Factions was any indication of article growth, we should accumulate another 3,000 or more new pages once Nightfall goes live, so we should scoot up that ranking pretty quickly over the next month or so. --- Barek (talk • contribs) - 16:37, 26 October 2006 (CDT)


 * I think one of the reasons that we have more page views than we might expect is because we have a lot of very small articles. Every skill in the game is a seperate article, and I know it's not very difficult for people to link to the GuildWiki numerous times when they're talking about a build. How many times have you gone to a build page and opened up some or all of the skills in seperate tabs? I do that now and then when I'm trying to decide which build to use.  &lt;LordBiro&gt;/&lt;Talk&gt; 05:22, 27 October 2006 (CDT)
 * True, but it's still an awesome figure. We rock! :) ~ Nilles (chat) 06:21, 27 October 2006 (CDT)

display party size more prominent?
Nightfall is out and several mission and quest descriptions are appearing in the wiki. What I personally miss most is the clear and obvious specification of the party size and the requirements of heroes for that mission or quest (to be honest I could puke each time I am in the game, form a party of human players of max party size and then can't enter the mission because I have to replace a human for a stupid hero again without being told about that requirement BEFOREHAND). My suggestion is to put a note somewhere on the very top of each mission and quest descriptions which says something like "8 player mission, hero Koss required". Maybe even at the mission and quest overview Mission overviews (Nightfall) there could be some small indication like "Venta Cemetery (8 players incl. Koss)". T.T.H. 02:47, 30 October 2006 (CST)
 * Not to sound like an ass, but it usually tells you in the game if a hero is needed for a mission before you try to enter it. That said I don't think that is any reason not for us to mention it too.  --Rainith 03:11, 30 October 2006 (CST)
 * Exactly, "before you try to enter it". At this point of time I usually have organized a full, balanced party based on the available slots in my party window. And then I have to kick somebody again because "before you try to enter" it tells me "nope, you need Koss, go kick somebody". This whole thing is about "comfort" the guild wiki could provide and users don't need to try starting the mission - which is actually quite tricky in NF because even missions are starting with questgivers and the very flexible amount of "yes"-"yeah"-"sure we are ready"-"goddammitgonow" you have to click through. I say the wiki could improve its service by providing that information - more clearly than the game itself does. T.T.H. 02:37, 31 October 2006 (CST)
 * Ofcourse the wiki will have thet information. Feel free to edit the articles to include the limitations/requirements. --[[Image:Gem-icon-sm.png]] (talk) 02:53, 31 October 2006 (CST)
 * Actually TTH, what I meant was that if you watch the cutscenes and read what people are saying around you, you will be given a good idea of who has to come. --Rainith 06:24, 31 October 2006 (CST)

Rainith, I find your comment both rude and illogical. Your argument is that this information should not be included on the Wiki because it is available in game. Guess what... almost ALL the information on the Wiki is available in game. With the exception of superfluous information such as mission walkthroughs and builds. Skills, Quests, Items, all of this information is available in game, and according to your logic should just be deleted. Your argument is laughable at best "If you watch cutscenes and read the info around you" Well if you look at the skill in game you also don't need the information on that skill posted online. TTH's suggestion is quite valid, and all possible information about everything should be included on every page of the Wiki. That's the POINT of the Wiki. The question remains, is TTH merely suggested this, or has he taken a proactive stance in updating this information on the various mission and quest pages as he comes across the information in game? ErkDog 12:19, 6 November 2006 (CST)
 * ErkDog, I think you need to read Rainith's post more carefully and owe him an apology. Rainith specifically agreed that such information should be added (see his very first response above). Rainith was just clarifying about being able to deduce that information before attempting to start the mission. -- Ab.Er.Rant (msg Aberrant80) 04:36, 7 November 2006 (CST)


 * Actually, you are right, based on the way he phrased the first part of his sentence, due to the overall tone, "Not to sound like an ass..." I must have missed the double negative in the second part of his statement. "I don't think we should not include it."  I missed the second not.  Moral of the story, don't use double negatives, especially when you start your thoughts off in a negative manner, it's easy for people to assume the rest of your statement is negative.  Although in this case, the second part of his statement was in fact negative, dually negative, lol.  I stand corrected.  ErkDog 07:03, 7 November 2006 (CST)


 * So far I only suggested this. I don't feel "VIP-ish" enough to edit such important pages like Nightfall Mission Overview just out of a personal idea. In addition I didn't "care" the last weeks due to the wiki's "unresponsiveness". Back on topic: While playing with friends during the last week I once more started to absolutely HATE HATE HATE the varying party size based on hero requirements. It forces you to kick friends out of the party or it hinders you from taking friends into your party because you know you have to kick them again after the next quest. If that very mission overview would state short and precise number of players plus the required heroes' names it would help during party organization (sorry if I repeat myself but that issue is my personal biggest annoyance in the complete Nightfall campaign). P.S.: calm down people, it's just a suggestion, and I never felt offended myself. T.T.H. 04:21, 16 November 2006 (CST)

Create "Skill:" namespace ?
I think we should create "Skill:" namespace. We do have a lot skill sections now outside regular skills, elite skills, monster we now have also title skills and ride skills for pve only. A reorganization under a form similar with what has been done with the builds sections will allow faster and easier access to them for everybody.--Phoenix 20:34, 1 November 2006 (CST)


 * I do not comprehend your vision of how creating the Skill: namespace would improve/change anything. - 21:20, 1 November 2006 (CST)


 * I think it would only add confusion, not make stuff easier. --[[Image:Gem-icon-sm.png]] (talk) 06:33, 2 November 2006 (CST)


 * I don't agree, a Skills namespace does not solve anything. By this logic we should have an item namespace, and a location namespace. The only reason for having builds in their own namespace is because they are fundamentally different to other kinds of article, i.e they deal predominantly with opinions rather than facts.  &lt;LordBiro&gt;/&lt;Talk&gt; 09:47, 2 November 2006 (CST)

What about translation?
Don't know if this has been discussed yet, but what do you think about translating guildwiki in other languages? Me, and hopefully some other, could care about the italian version.

What do you think about this?
 * We don't want to host other languages here, but everything is CC 2.0 by-nc-sa so you can start up guildwiki.it based on our content if you want. --Fyren 05:44, 2 November 2006 (CST)

Server stuff
A couple days ago Gravewit gave me root access to the servers. I've poked around and made some changes but no amount of effort is going to solve the real problem: the machine running the web server simply lacks the CPU power to handle the amount of traffic we get. There's a new machine on the way but it'll be at least a week plus time for us to get things set up. In the short term, I'm going to try out some caching options which may or may not help (and the best case probably isn't even going to be a substantial improvement). This means in a few hours, I guess from 3-4 AM EST, the server will be going up and down as I do things. Do we want to put a message in MediaWiki:Sitenotice about it? I ask because saying "the wiki might be slow or not respond at all" is sort of the general state of affairs. --Fyren 22:13, 3 November 2006 (CST)


 * Yes, that is the courteous and responsible thing to do. To say nothing is to tell people to expect us to suck. --Karlos 22:31, 3 November 2006 (CST)
 * Thanks for the update. :)  I'll post this info on the GWG forum too.  --Rainith 22:49, 3 November 2006 (CST)
 * (Edit conflict)Not that everyone needs to comment on this but I agree with Karlos, as much prior notice as possible should be given. Also congrats Fyren, it's good news for all to have another person such as yourself working on the servers even if your first act is apparently to rip them down :P --Xasxas256 22:52, 3 November 2006 (CST)

Feel free to change the formatting of the notice if any (admin) cares. I'll be testing things out as much as I can before the mentioned timeframe. I don't expect anything I do will impact the server before I inflict my experiment on the wiki at large... but I tacked on the second sentence anyway. --Fyren 23:03, 3 November 2006 (CST)


 * I commented at MediaWiki talk:Sitenotice but I'll say it here instead. Couldn't we instead say
 * The GuildWiki will be down intermittently from 3-4 AM EST (8-9 AM GMT). We will be making some improvements to our servers during that time.
 * You could also include something about the new server although maybe not until we have firmer knowledge of when it'll arrive. --Xasxas256 23:12, 3 November 2006 (CST)
 * I don't want to sound optimistic about what I'm doing tonight since I'm not. It's something I've suggested to Gravewit all along but, as I mentioned, right now it can't help much in the best case.  From my perspective it's more to test things out so now it'll take less effort later when we do get the new machine.  --Fyren 23:23, 3 November 2006 (CST)
 * Well we appreciate it nevertheless. An admin couldn't change the notice though could they? I find the time thing particularly confusing. How about:
 * The GuildWiki will be down intermittently from 3-4 AM EST (8-9 AM GMT). We will be making some improvements to our servers during that time in preperation for a new server which should be up and running in the next fortnight.
 * It's pretty similar to my first suggestion but mentions the new server without giving a strict deadline on when it'll be up. I personally think the first message I suggested is fine, it doesn't make an real promises, it just says "improvements". The second one is better if we want it to be known to all that a new server is going in... --Xasxas256 23:33, 3 November 2006 (CST)
 * Actually timezone wise it'd be better to use CST and GMT (as that's our timezone on the server currently, look at sigs). Ie 2-3 AM CST (8-9 AM GMT) in the message. It's probably worth saying that it's happening on Saturday 4th of November too... --Xasxas256 23:50, 3 November 2006 (CST)


 * I made a slight change to the wording. Hopefully that helps - let me know if not. --- Barek (talk • contribs) - 00:05, 4 November 2006 (CST)

Everything seems to be working as intended now. Does responsiveness seem any different? --Fyren 02:40, 4 November 2006 (CST)
 * Response time seems better, but that could just be because it is 1am - 4am in most of the US. --Rainith 02:59, 4 November 2006 (CST)

Seems somewhat better. I'm sure you guys know what you are doing, but there are literally dozens of things you could do to alleviate load. Based on the nature of httpd and mysql servers, I find it difficult to believe that the bottleneck is the httpd server. Especially with only 50 requests per second. (I think I saw that somewhere else on the wiki.) If it is actually the httpd server overloaded, then all you have to do is add another httpd server to the cluster and pop in another DNS record for load balancing. Which would take merely a few hours. I am unsure why you indicate that it takes days to setup a server. I've never spent more than a few hours setting up and configuring a server. Especially not a single purpose server such as one that will serve only httpd requests. ErkDog 21:27, 4 November 2006 (CST)
 * At peak a lot more than 50/s. That wouldn't even be much if we were serving static pages instead of pages generated by MW. MW is pretty slow (really slow compared to a normal CMS).  As for the new machine, I didn't say "days" for setup, I said "plus time for us to get things set up."  We don't have the hardware yet.  --Fyren 02:11, 5 November 2006 (CST)

Right I think you seem to have validated my point... MW is not inherently slow itself. Something somewhere is waiting on something. IE, MySQL. There are several things that could be done to the MySQL server to speed it up as well. Can the contents of the my.cnf file as well as the system specs be provided for comment and suggestion? Also, the database structure itself? The types of tables, etc... For example all tables should be MyISAM instead of InnoDB if you are using InnoDB. InnoDB incurs an approximate 400% overhead. CRAAAZY I know!! ErkDog 08:03, 5 November 2006 (CST)

Can you provide the output of iostat -x 10 5 on both the httpd and MySQL servers? Can you provide the output of uptime for both servers? --ErkDog 08:09, 5 November 2006 (CST)
 * MW is inherently slow itself. The hard drives are doing fine.  The DB machine CPU load is practically non-existent and the apache machine's load is high because of MW.  Normal page render times for MW can be hundreds of milliseconds real time for a decent CPU that's not already under heavy load.  MyISAM is not suited for large tables under heavy use and performs better than InnoDB for the opposite.  --Fyren 15:03, 5 November 2006 (CST)

Fyren, thank you for your response. For my edification would you mind pasting the output of uptime and iostat -x 10 5 on the DB server? As well as the httpd server?

I am also confused about your reply re: MyISAM versus InnoDB. You provide some theoretical information regarding which is best used when, but have not provided information as to which type of tables GameWikis is in fact using.

I assume this is because you believe the table type to be obviated by the fact that the MySQL server (as reported by yourself) is not under a significant load.

Can you please provide the uptime and iostat -x 10 5 outputs of the two servers in question? If you are uncomfortable posting this information on the Wiki itself, by all means please e-mail it to me erkdog *at* fiftypounds.com

I would very much like the opportunity to apply my server administration expertise to gamewikis in an effort to provide a more snappy services for all. Please do not think I am trying to undermine your efforts, or that I am questioning your ability to administer servers. But another pair of eyes never hurts.

I would also very much like the opportunity to analyze the structure of the database and provide insight on areas which may benefit from a different table type, or may benefit from use of an index or two.

Thanks ErkDog 12:11, 6 November 2006 (CST)

Still mostly twiddling my thumbs and waiting for the hardware which was supposed to have been delivered and set up for us on Thursday. I haven't been able to do much, as I've said. The random ban issues and SQL errors about duplicate entries in the cache may or may not be fixed (it's hard to tell with the wiki crawling). I'd like to hear if anyone still gets either. To Erkdog, I appreciate the intent, but I'm not going to act as your shell so you can go through the same things I did when Gravewit gave me root access just before I created this section. --Fyren 23:48, 12 November 2006 (CST)


 * Thank you for your response albeit significantly delayed. I think you fail to realize that if you posted this information in a public place it could be reviewed by several people, not just myself. None of it is sensitive nor does it propose any kind of "threat" for you to provide it.  Failure to provide it is merely counter productive.  It took you longer to draft your response saying you would NOT provide it, than had you merely copy and pasted the requested information. ErkDog 07:01, 13 November 2006 (CST)

Just to let you know: This is the avatar of one regular user on the GWGuru forums. No comment. -- 03:12, 14 November 2006 (CST)


 * Regardless of how personally I take that, it's still pretty funny :P  &lt;LordBiro&gt;/&lt;Talk&gt; 04:20, 14 November 2006 (CST)

Lightning fast
Looks like the new hardware did cut it. The server's response is lightning fast at the moment, and I didn't notice a single bug so far. Thanks, good work, Fryen and Gravewit! (even though the problem shouldn't have come up in the first place, and solving it took a bit longer than we were all hoping for. :p ;)) Anyway ... it's water under the bridge, and GuildWiki is back in business. I think you can remove the "GuildWiki is experiencing a lot of slowness ..." sitenotice now. :) -- 05:44, 16 November 2006 (CST)
 * Yep seems very good so far, well done to those involved, hopefully we can maintain this kind of server performance. --Xasxas256 05:56, 16 November 2006 (CST)


 * I love the new server. ^_^ Arshay Duskbrow 17:20, 16 November 2006 (CST)

Guild Wars concept art
I found it on this site

http://www.conceptart.org/forums/showthread.php?t=74118

could it be added to the wiki in any way? and there's ALOT


 * Not without written approval from ANet. It looks like the person who posted that was from ANet's art dept and he had to get approval to post it also.  --Rainith 00:31, 4 November 2006 (CST)


 * "Tinfoil", the guy who started the thread, is Daniel Dociu, ANet's head art director. A link to that thread has been in his article for quite a while. I don't think the art should be added to GuildWiki though, for bandwidth reason, and because a wiki isn't a good place for such a gallery. --[[Image:TurningL sml.gif|Tetris L]] 07:11, 16 November 2006 (CST)

Servers slower than a [slow thing]
Is it just me, or are servers obscenely slow or have been obscenely slow recently? It's downright rediculous. I know it's not me since a 70 meg Crysis trailer downloads in under half a minute for me. Anyone else noticing this too? --Mgrinshpon 17:46, 5 November 2006 (CST)


 * Lol. The server isn't the only thing that's slow. j/k. Look up two headings. &mdash; Gares 18:19, 5 November 2006 (CST)
 * I've noticed the same and so have other fan sites as well (I'm on a cable connection at 2 different locations and it's the same). I'm seeing people go to other sources now because it's so slow to load.  What gives?  Does anyone know how to fix this?  I barely visit these days too since it takes so darn long to load and 1/2 the time I get a 504 error timeout code page.  This is getting ridiculous.  HELP!--[[Image:VallenIconwhitesmall.JPG]]  Vallen Frostweaver  08:48, 10 November 2006 (CST)


 * GameWikis refuses to take advantage of numerous resources and opportunities to deliver their content to the community responsively. I have requested in multiple places information on system load to put another pair of eyes on their issues.  Additionally I have requested in multiple places to lias with someone regarding the automation or even manual creation of a mirror.  For example, the Wiki has basically been unavailable for almost a week now.  If, a week ago, my communications had not been dismissed there could be a 100% up to date complete mirror of GuildWiki for people to access.  Furthermore, if the information I requested was provided it's entirely plausible that a configuration change could be suggested which also resulted in more performance.  But GameWikis refuses to explore either of these possibilities.  Why?  Your guess is as good as mine.  ErkDog 09:42, 10 November 2006 (CST)
 * Any idea what I can do to help petition or inform others about this needing a fix? I don't know where to go to voice my concern over this beyond this spot for now.--[[Image:VallenIconwhitesmall.JPG]]  Vallen Frostweaver  11:03, 10 November 2006 (CST)

As far as I know this is the most appropriate place to discuss it. You can see above that I requested information, which was vaguely responded to by the current GameWiki server admin, but then beyond his vague response has gone unanswered. You can also find discussion of a mirror system on this page, which also remains uncommented on by GameWiki staff. I remain perplexed as to why nobody on the GameWiki staff cares enough to comment on the issues, much less to actually proceed forward with them. Heaven forbid the community benefit from assistance in delivering content or from discussion of server administration practices. It defies my comprehension. ErkDog 12:52, 10 November 2006 (CST)

Skill Data / Progressions
Apparently a lot more skills need to be fixed. A comparison has been made and apparently here is a small list with the mistakes on GuildWiki: http://gwshack.us/forums/viewtopic.php?t=357. --Phoenix 06:24, 6 November 2006 (CST)

Mirror
Take a look at this and this. I think it is a great idea as a backup whenever th wiki is slow or down, but it would be great if it could somehow be institutionalised so that guy down not have to use a DL program to get the wiki pages. --Xeeron 06:41, 6 November 2006 (CST)


 * Yeah, I outlined what is required to provide an effective and complete mirror. Basically all I would need is access to a snapshot the PHP / images and an almost complete SQL dump.


 * Perhaps Fryen would be willing to provide this? If nothing else it would be a stopgap measure until new hardware was in place, and instantly provide a complete mirror. You could put a temporary link at the top of the page to indicate if people are going to be reading the Wiki only to my server.


 * However, I think it would behoove GuildWiki to strongly consider setting up the infrastructure required to provide permanent mirrors. Fryen, I would be happy to discuss this with you in detail as to what is required to keep a regularly updated Mirror online.  It can easily be done without creating significant load on the site's servers.  I guarantee you that my mirror software continuously crawling the site generates more resource use than providing a daily snapshot of the PHP / IMG and salient SQL tables.


 * The database dumps can be tailored to the point that they are not large. The main content of the articles would be the only thing necessary for a mirror.

History is not necessary. Incremental updates to the database can also be created via binlogs.


 * Not to mention the fact that after 4 days of crawling I still am nowhere near a complete mirror.


 * If nothing else, a complete DB Dump and PHP / IMG snapshot could be provided so I could create a complete mirror in the interim while you wait for new hardware.


 * If the information I requested above in the server section was provided I could also provide some insight into the slowness and make suggestions on configuration that would speed things up as well. ErkDog 11:53, 6 November 2006 (CST)

Gravewit had apparently already wanted to figure out how to get dumps out to people but not made an effort (possibly due to concerns about bandwidth). Dumps of the current article revisions are easy to make (and small) but I'm not sure if we even want to attempt to provide our images as an "open" download since even only the current versions total 1.3 gigs. I will get something regular going for at least article dumps but only at some point after the new hardware is around. --Fyren 23:48, 12 November 2006 (CST)


 * Fyren, A compressed version of the PHP and IMGs would be nowhere NEAR 1.3 Gigs. Additionally you would not have to provide the mirror content "publicly."  Assuming the database and images themselves were even as large as 2 Gigs (Which they are not.)  That should be insignificant toward your monthly transfer limit.  If it is not, perhaps you should consider an alternative hosting provider. Although the size of the data is irrelevant if you only allow mirror peers to access it.  Rar should turn about 1.3 Gigs of images into a few hundred megs at most. Also, there are HUNDREDS of places you can put the dumps that provide free download.  RapidShare, File Planet, Planet Miror, just to name a few.  You of all people should know this being a server administrator.  You should also know that even gunzip would squish the hell out of those images, much less a sophisticated archiver such as rar. ErkDog 07:06, 13 November 2006 (CST)


 * Firstly, I'm not a server admin.


 * "You should also know that even gunzip would squish the hell out of those images, much less a sophisticated archiver such as rar." On the subject of image compression I think you will be hard pushed to find any algorithm that will compress 1.3 gigs of predominantly JPEGs into an archive weighing in anywhere below 1 gigabyte.


 * As far as I'm aware, common lossless compression algorithms (including RAR) build a table of repetitive patterns of bytes in the input and replace those repetitive patterns with shorter patterns. The most commonly recurring patterns in the input are replaced by shorter patterns in the output. JPEGs are reduced in size using a lossy algorithm, and then further reduced in size using this lossless algorithm. Therefore there are very few repeating patterns in JPEGs and therefore less to compress, so to speak.


 * The more JPEGs you try to compress the greater the level of compression you will (typically) be able to achieve, since more JPEGs equals a greater chance of repetition. Bearing this in mind I just carried out a test, compressing 1.82 GB of JPEGs that I happened to have at my disposal using WinRAR's "best" compression option. I managed to produce an archive of 1.71 GB.


 * I used 1,184 JPEGs, so some basic maths tells me that they average at about 1.57 MB each. I presume that the average size of images on the GuildWiki will be smaller than this, so I would imagine the images from the GuildWiki would compress better than my images. If the GuildWiki images did compress at the same ratio as mine then they would produce a RAR weighing in at about 1.22 GB.


 * I am by no means an expert on compression algorithms, so if my information here is incorrect please let me know.  &lt;LordBiro&gt;/&lt;Talk&gt; 18:43, 13 November 2006 (CST)


 * Just as a data point, the bzip2'd compressed database, for the current wikipedia articles, is around 1.6GB (you can see this by going to the wikipedia database download pages -- yes, unlike guildwiki, the much larger wikipedia does allow database downloads!). As this wikipedia dump has over 1.4 million articles, I think it's safe to say that a compressed guildwiki dump would be much smaller.  Reference: http://en.wikipedia.org/wiki/Wikipedia:Download  And, given that there are probably a number of places who'd be willing to host periodic guildwiki current page dumps, I don't think bandwidth is really an issue.  Frankly, I've love to have an offline-accessible guildwiki setup (xampp and others are your friends).  21:37, 13 November 2006 (MST)
 * I did say a dump of the article text would be small. You'll also note that the Wikimedia foundation has just a few more servers and a little more resources in general than us.  --Fyren 23:14, 13 November 2006 (CST)


 * As Fyren said, he did say that a zipped version of the articles would be small, and of course I didn't say anything otherwise in my post. And just to demonstrate the difference in size and capacity between GuildWiki with its 2 servers and MediaWiki with its... well... several hundred servers, please see this pic.  &lt;LordBiro&gt;/&lt;Talk&gt; 04:17, 14 November 2006 (CST)

Yeah I forgot JPGs can't be significantly "squished" but the fact remains, assuming the salient database data and PHP / JPG data are as large as 2 Gigs (Which they would not be) it would NOT create significant pressure to create backups and allow mirror peers to download them. When you consider the fact that the mirror content can be offloaded to an entirely different place the size utterly becomes a moot point. Then when you consider the fact that thousands of db queries and http requests would be offloaded to the mirrors, the relatively tiny amount of resources required to create a backup and provide it to mirror peers FAR outweighs the few resources required to create and distribute the mirrorable content. We should stop discussing whether or not you have the resources to supply mirrors. It would be painfully clear to a retarded tree frog that mirrors make sense. What we need to do now is discuss a way to make it happen:

1) Create a script to dump the database with the exception of the user table and some of the more frivilous data. 2) Create a script to compress the img content into an archive. 3) Create a script to compress the PHP content with the exception of config.php or any other files which contain GW's actual DB credentials. 4) Create an HTTP folder which is protected by an .htaccess file to only allow download from specific IPs.

On the mirror end.

1) Create a script to grab the dumps on a regular basis 2) Wipe and import the new database. 3) Uncompress the IMG and PHP archives.

Viola mirror.

Fyren, let me know if you need help creating the scripts or .htaccess file above.

Thanks, ErkDog 07:44, 14 November 2006 (CST)


 * Fyren, I just had an epiphany. Something that would be completely transparent to gamewikis and require little to no work on your end.  Squid (or another cache).  Instead of people setting up mirrors of the actual content.  People could donate Squid (or another cache) processes.


 * This could be done completely transparant of the existing servers and setup. Basically what we would do is come up with a good squid (or another cache) that obviously listens on port 80.  We setup another DNS (squid.gw.gamewikis.org) for instance, which would be an RR DNS to all available squid caches.  And viola.  Anybody using that host hits the squid servers and potentially avoids a query to the main Guild Wiki server entirely.


 * Down the road it could be modified such that something monitored the status of the configure cache servers and (as appropriate) deleted or added their DNS records to remove down caches from the RR DNS.


 * I have a bit of experience with Squid but am certainly not a Squid master. I know if nothing else, this will definitely cache image requests.  It may be necessary to remove the counter at the bottom of -every- page.  Something that should have been done already anyway considering the excessive load the site is under at the moment.  That's an extra data base query and update on every page load.  Eliminating that Update on EVERY page load should save a few cpu cycles and I/O definitely.


 * Let me know what you think. ErkDog 10:49, 14 November 2006 (CST)


 * We're already running a local squid and having (only) more squids elsewhere won't help the problem of apache/MW load besides that (at best) we could not run squid here. Once whatever the hell is happening with the new machine gets sorted out, we'll have squid on its own machine and squids elsewhere won't help besides for bandwidth.
 * Also, with "remote" squids, a cache miss would mean things will be even slower for users as squid ends up going to our apache. You'd ideally want an apache server local to the squid... and then you'd also want a local DB slave/mirror... and maybe memcached (making it a 2-3 machine setup, even if those machines aren't dedicated to us).  It's a lot to ask for on top of bandwidth.  I'd also be generally apprehensive about trusting machines in someone elses' care, but that's a separate issue.
 * About the page view counts, I had disabled them briefly to gauge the effect. It made no apparent difference.  --Fyren 19:54, 14 November 2006 (CST)

Memcached would help more than Squid does. As for remote squids, any thing that can be served by the squid is 100% load relieved from your Apache / Mysql. Memcached can also be run on the same box as apache. Memcached is designed more to relieve MySQL Load though. And according to you it's HTTPD Load not MySQL that needs to be relieved. If you don't disable the page counters then Squid can't really do it's job. Every page load is different and it can't cache any of them. The reason you didn't notice a difference is because the load is so ridiculously high it made a relatively inconsequential difference. But you are preventing squid from doing it's job by not leaving it on. Futhermore, it is DEFINITELY an update on ever page load. ALWAYS. You should turn it off. It's frivilous information, breaks squid, and increases I/O load a great deal. ErkDog 21:16, 14 November 2006 (CST)


 * It doesn't break squid unless by "break" you mean "users seeing a cached page will see a cached number." Incrementing the count when a request goes through to apache doesn't cause squid to purge the page from its cache.  --Fyren 21:30, 14 November 2006 (CST)

Well, I guess it would depend on how you have squid configured, if it's configured to IGNORE IMS then yes it won't matter. But it also won't pick up LEGITIMATE changes either. Look at the hit % it can't be that great if you don't have Ignore IMS and w/o disabling that and counter ErkDog 22:28, 14 November 2006 (CST)
 * MW tells squid to purge pages when appriopriate. MW knows when a page is edited/a file is uploaded/whatever.  --Fyren 23:15, 14 November 2006 (CST)


 * O M G whatever, I'm done trying to help with this crap. Squid has no way for external programs to notify it when it should expire pages.  Squid operates off IMS or a hard timeout depending on how you have it configured.  MW cannot tell Squid to expire a page. Fyren > * good luck.  I'll stick to updating then Wiki when it bothers to be responsive enough for me to do so. ErkDog 23:19, 14 November 2006 (CST)


 * Don't let the door hit you on the way out. --Fyren 23:37, 14 November 2006 (CST)


 * Fyren, he was just trying to help. No need to rub it in and alienate him. :( ErkDog, I appreciate your attempt to help, but I'm also confident that our server is in very good hands with Fyren. We'll see if the new server hardware gets our performance back to normal. I take it, if GWiki still runs sluggish after the server upgrade, we'll reconsider mirror options. --[[Image:TurningL sml.gif|Tetris L]] 02:22, 15 November 2006 (CST)


 * To be honest, Tetris, Fyren didn't say anything unreasonable until after ErkDog had thrown a tantrum. I wish Fyren hadn't just said that, but I still don't think he was in the wrong. It's a very stressful time at the moment for Fyren, having to mediate between an angry wiki and Gravewit. I know ErkDog is/was only trying to help, but Fyren has not criticised any of ErkDog's suggestions, he's only explained why certain suggestions would not work, and what he is capable of doing.


 * Fyren might not have all the answers, but from my conversations with him I'm confident he knows a lot about MediaWiki and the technologies surrounding it.  &lt;LordBiro&gt;/&lt;Talk&gt; 03:00, 15 November 2006 (CST)


 * It was kinda nice to see some discussion unfold really, talking it out is good not just for drinking problems but wiki problems too. Anyway it's a shame ErkDog has departed, it was all good until his an Fyren's last post. I'm not so sure about the "responsive enough" part, Fyren doesn't speak for all of us. That said I'm pleased that we're being kept in the loop as progress is being made. --Xasxas256 03:16, 15 November 2006 (CST)


 * Just trying to do some damage control here. I don't really blame Fyren, I can understand his reaction. It wasn't exactly totally uncalled for. Actually, ErkDog's comments towards GuildWiki on forums over the last few days were more and more hostile. Fyren obviously knows his ins and outs as a server admin, and is not in urgent need for advice. (He's in need for better hardware. ;)) But - like Xasxas said - discussing proplems and possible solutions in public can never be wrong. Many of out users work in IT as server admins and there may be people among them who's advice is definetly worth listening to, and turning down people who try to help gives the wrong message.
 * I've read various forums a lot over the last two weeks (wasting my time away as the wiki is down ;)), and the amount of negative comments I've read about GuildWiki's responsivenes really worries me. Many of our users do not read the GWiki talk pages (even when the wiki is fast, let along when it's creeping slow or not accessible at all). They only see the sitenotice (which doesn't tell much and hasn't changed in almost two weeks), and the little bits of information on the GWGuru news and forums (which are not very informative and positive either). I trust that behind the scenes Fyren and Gravewit are working hard to solve the problem, but the public doesn't see that. To our average user it looks like we're sitting on our asses, not doing much about the problems, and even have the nerves to turn down offers for help. All in all, our PR is LOUSY at the moment!! That's why I asked for a bit more transparency, via the site notice, via our GWGuru forum, and via the gamewikis blog. Maybe Gravewit should ask Gaile Gray to teach him a bit about "community relations". :( /sarcasm --[[Image:TurningL sml.gif|Tetris L]] 04:28, 15 November 2006 (CST)


 * Yeah Tetris, I agree with you on the PR front. People don't read talk pages, especially not when the server is as slow as it is at present. I think it would make a lot of sense to put up a notice explaining the current situation.  &lt;LordBiro&gt;/&lt;Talk&gt; 06:41, 15 November 2006 (CST)

Well squid obviously isn't doing it's goddamn job then. If you claim it has all the pages cached and keeps them until MW tells it to purge them. Otherwise the Wiki would NOT be ridiculously slow, and Squid would be taking the large majority of the page hits, obviating the load from HTTP and the Database. HTTP would not be "overloaded" There are NOT that many updates that should be kicking pages out of Squid. And I stand corrected on the purge. It wouldn't surprise me if Squid is on the same damn server as the HTTP. But I wouldn't know because Fyren refuses to answer any questions about the overall configuration of the damn wiki servers. If the box w/ HTTP and Squid on it is overloaded, and the MySQL server not under load, how much sense would it make to have moved Squid over to the MySQL Box? But I find it difficult to believe that the Squid server was so pounded that it can't serve pages. Especially since Fyren indicated specifically that HTTP was overloaded. But by all means, GameWiki should continue to refuse to share configuration information and let it's Wiki remain slow as balls. If Fyren meant to say that the SERVER with http and squid on it was overloaded, then that's what he should have said. But Squid clearly isn't taking the load off the HTTP it should be. Serving static pages is not resource intensive. Fyren claims that MediaWiki is slow. That become irrelevant when a Squid process is serving a cached page. I also find it difficult to believe that that counter doesn't kick pages out of the cache. That counter is a part of the page, and it being updated should and most likely does tell squid to kick the page. Turn the damn counter off, it's frivolous and even if it DOESNT kick pages out of the cache it's extra database queries for no reason on -every- friggin page. Given my post Fyren's response was not unexpected. I don't really care about that. What pisses me off is that nobody will share any goddamn information about the Wiki. This entire conversation has been nothing but vague information and horse crap. Nobody can have a productive conversation about anything relating the servers or mirrors because nobody will provide any friggin information about anything. Everything is met with vague responses or no response at all. Fyren will argue adamantly that Squid can be told to purge a page. But won't bother to comment on the Mirror system I laid out. Good thing that the GW staff takes the time to talk about the -important- things. Whatever. ErkDog 06:58, 15 November 2006 (CST)


 * Lol Tetris, as outlined above, most of the things I've mentioned here have gone either Ignored, or been sidestepped with vague answers. The first round of posts I made here ON THE TALK PAGE, went completely unanswered for over a week.  Like I said, it's good that Fyren can take the time to wax over the configuration of squid, but not elaborate on useful things that could serve as a catalyst to improving the wiki's performance.  One of those pages could have been referenced far earlier into the conversation instead of letting it progress to the point it did.  But it's really moot, had the conversation taken a more appropriate path, squid would have probably never even came up.  Since you guys seem so reluctant to setup real mirrors, and since you also seem to not discuss it worth a crap, I was merely brainstorming.  And Fyren hopped on one obfuscated point rather than hold a productive conversation about mirrors or server configuration. ErkDog 07:09, 15 November 2006 (CST)


 * I've just been browsing through this and burst out laughing. ErkDog, get a hold of yourself.  You aren't the only person in the world that knows a thing or two about server hosting.  Yes, a fresh set of eyes may help, but it should have been painfully obvious by your second message that Fyren does not need or want your help.  The fact that you keep coming back and then throw a tantrum when you don't get your way is just astounding. 68.209.230.87 03:48, 16 November 2006 (CST)


 * Lol dood whatever. It has nothing to do w/ getting any body's way.  100% of my annoyance comes from lack of communication on the GWiki staff or their lack of contribution of salient information.

I can't share the current server setup, as it's currently been in flux, and my information is out of date.

I can't tell you how transparent we're going to be, because I don't know. I can tell you that Fyren has significantly helped, in that a server admin is now discussing things with the wiki in general, even in a limited fashion.

However, I can tell you that Squid does not work the way you think it does. For example, Squid is not serving any of the requests that you or I submit, because we're logged in. If Squid served the pages, you wouldn't get that nice link to your userpage at the top, and you wouldn't get notifications that people have edited your talk page. Thus, Squid is not serving the overwhelming majority of pages here, as you think it does.

Squid is not the magic bullet for GuildWiki. While we appreciate your attempts to help us, take solace that we've done quite a bit of research on Squid, memcached, and load balancing solutions, and are evaluating every possible solution for merit and cost. If we come across any specific implementations that we'd like advice on, we'll certainly ask the community, as we know quite a few talented administrators frequent the GuildWiki.

&mdash;Tanaric 23:41, 16 November 2006 (CST)


 * Between the fact there are a lot more anonymous users than logged in users and that most pages have at least a few images, the majority of requests are for "anonymous pages" and images. So by number of requests, squid is doing a lot of the work.  Unfortunately, that work is the easy work.  We were already using a crappier cache strategy for anonymous pages before squid, but rendering the pages for logged in users is where the the problem was.  While everything but the DB was on the same machine, using squid dropped the CPU load a lot, but the difference was like going from "incredible amounts of load" to only "way too much load."  --Fyren 00:01, 17 November 2006 (CST)

"Fake" pageviews cause server slow-down?
According to our stats we have more than 200 million pageviews. According to the list of largest wikis that's FAR more than ANY other wiki out there, including the largest wiki of all, English Wikipedia, which has more than 100 times as many articles and 150 times as many users as GuildWiki, but "only" 65 million pageviews.

I simply can't believe that. I take it, either the GuildWiki statistics are incorrect. Or, if they're not, then it would definetly be worth investigating the reason! Has it occured to anybody that this may be one of the reasons behind the mysterious slowness we're currently experiencing? Maybe "something" on the server is generating "fake" excrescent pageviews, eating resources?! Kinda like an internal DOS attack. I'm not an IT techie, and I know very little about MediaWiki, MySQL and Apache, so I've got no idea what exactly may be causing a "pageview multiplyer" effect. It's just a hunch. Maybe somebody with more knowledge than me could look into it? -- 02:35, 14 November 2006 (CST)
 * The view stats for enWP are just from whichever of their many servers was queried at the time, not a total. You'll note the stats indicate there have been more edits than views.  The WMF servers easily get millions of article views a day (this suggests enWP has been getting about 4.6m views a day on average lately).  Also, the reference nature of a lot of our content means we have a lot of small articles that often get quick looks and lots of clicking around (comparing skills, looking up item stats, viewing armor images/crafting info, etc.) whereas other wikis probably have longer articles you don't fly through in one sitting.  Our numbers are accurate but a straight comparison doesn't take everything into account.  --Fyren 03:22, 14 November 2006 (CST)
 * What you say makes sense, and checking the list it is very obvious that the pageview stats are bugged. But I still find it hard to believe that out of 120 wikis we are the only ones with correct pageview numbers. We beat ALL other wikis by number of views total, views per edit and views per article, and all except two by number of views per user. That smells very fishy. --[[Image:TurningL sml.gif|Tetris L]] 03:50, 14 November 2006 (CST)
 * A lot of those sites may just have the page view counters off, freezing the reported views at whatever they were at. I'm too lazy to look through the page history to see which have been changing and which haven't.  I noticed Wikitravel's hasn't (checked since they have a similar number of good articles and edits to us).  --Fyren 19:57, 14 November 2006 (CST)

Reboot
Changing a setting that'll require a server reboot to take effect. Probably doing it around 3 AM EST, shouldn't be any interruptions besides that. Dunno if you want me (or whoever) to put it in the site notice since it'll be minutes at most. --Fyren 23:22, 14 November 2006 (CST)
 * Don't bother. People have become so used to GWiki being down that they won't even notice. ;) --[[Image:TurningL sml.gif|Tetris L]] 01:55, 15 November 2006 (CST)
 * I'd like to say that if we're going to improve the GuildWiki, be more accountable and helpful to the community we should let them know. But at this stage I'd just go ahead with it. --Xasxas256 02:21, 15 November 2006 (CST)

Guild Wars Top 200 anyone?
Guild Wars Top 200 has quite a few sites in it, but no GuildWiki? While these kind of websites are 10 a penny we can't let them have a list without GuildWiki can we? Just out of interest, I accessed this link from GuildWarsGuru, I generally don't click these kind of things but when I noticed GuildWiki wasn't in it, I was appalled. --Jamie  05:54, 15 November 2006 (CST)


 * We don't need a measly site like that to tell us that we're Kind Of A Big Deal, do we? ;) Hell, how do they determine their rating anyway? Most of the sites they list are not listed on the official fansite listing, and vice versa. Their listing means absolutely NOTHING. Simply forget about them. --[[Image:TurningL sml.gif|Tetris L]] 07:22, 15 November 2006 (CST)


 * When people ask questions in-game, they are directed to the wiki. Nobody directs them to that gonzo listing, so I frankly couldn't care less whether we're on it. The measure of worth we're looking for here is actual use by players, and I think the wiki has succeeded in that. Kessel 22:38, 15 November 2006 (CST)