Guild Wars Wiki:Reporting wiki bugs/archive 9

RSS settings - Please read and give your thoughts!

Hi guys! I just got a call from my wiki IT compadre, and he mentioned a few concerns with our current RSS system. He'd like to address this before everyone is off for the holidays to keep the wiki as stable as possible when everyone is home with their families, so please give me your thoughts as soon as possible so I can pass them along to him. Anyways, here's where we're at:

As I mentioned before, excessive RSS queries were the reason for the wiki's downtime earlier last week. We've been able to track exactly how excessively people were using the feeds, and found out that people have their readers set to pull new data from the site as frequently as every 9 seconds! When hundreds or even thousands of people are hammering one page almost 400 times an hour a piece, it's no wonder there have been performance issues. In addition to this, our wiki load has increased significantly during the holidays, so there are a lot more people on the site. To keep the wiki from having further problems, we'd like to set up some reasonable restrictions on the RSS. We'd like your help to figure out what's a fair limit, since you guys are using the feed the most.

I know that cutting the feeds to update too infrequently kinda destroys the point of a RC RSS, especially since a lot can happen on a wiki in 15 minutes. I also wonder if some of you use the feeds as an easier way to patrol for vandals? Even with these points in mind, we definitely need to do something to make the hit on the feeds more manageable because we don't want to have downtime because of something like this.

If possible, we'd like to set the refresh limit somewhere in the range of 4 - 6 times an hour (IE RSS feeds update every 20 minutes by default). We'd also be able to set how many results are returned for each refresh, so we'd be able to say something like "push 60 of the most recent changes every 10 minutes". We're open to suggestions, so please let me know what limits you guys think are reasonable.

Thanks in advance :) -- Emily Diehl (talk) 22:20, 22 December 2008 (UTC)

"our wiki load has increased significantly during the holidays" - sorry, that was Wikichu :P

I tried to use the RSS feeds a long while ago but stopped using them within some hours as all feed readers I tested didn't received multiple entries with the same page id (=multiple changes on one page), so it was quite unuseful for watching for RC changes... That's why I decided to go back to the HTML version :P Much better and gives more information when you need it. (What I want to say, I wouldn't mind disabling the RSS at all.. but I looking at your numbers there seem to be some people who actually use it). poke | talk 22:33, 22 December 2008 (UTC)

I, like Poke, don't really see the use in an RSS feed for recentchanges in the first place, let alone one hitting the site every 9 seconds (and this comes from a HUGE fan of RSS feeds). Certainly I'd have no problems with limiting the rates for it as much as needed to ensure stable general wiki browsing, as that should always be the primary concern. - Tanetris 22:48, 22 December 2008 (UTC)

HTML version is better as then you don't have to watch your rss constantly then open your browser/new tab and make the changes. Much easier to just be right there. I also don't use the RSS as the html is just better by design. Dominator Matrix 22:52, 22 December 2008 (UTC)

I've only been using it for a short while now (just before it went blank), so I can't really comment on a good time for refreshes. I use Google Reader and haven't changed my settings yet (I don't think that's even possible). I understand too many RSS feeds can be a problem, so I would understand if it's disabled, but limiting while keeping it usable is fine too. Perhaps talking to other Wiki admins (for instance wikipedia) might give a good idea? I use the feed as an alternative to the watch list. I don't want to be alerted on Talk page edits, so I filter the RSS feed, that's a lot harder on the HTML version. Cavalier 23:50, 22 December 2008 (UTC)

Can't we just do this Amish-style and simply press F5 or the RC link every time we want to refresh the page? I've been doing that the past eighteen months and its worked fine for me. calor (talk) 17:45, 23 December 2008 (UTC)

I think an RSS update time of 10 or 15 minutes is entirely reasonable. Limiting the number of items in the feed would also be sensible. On the server-side, worth using Wikidiff2 - better than the original wikidiff used by the RC page. Me55enger 23:01, 30 December 2008 (UTC)

(Reset indent) Hi guys! The team has come up with a pretty slick solution to the RSS issue. We've placed the feeds on a separate subdomain on guildwars.com, so anyone monitoring wiki feeds will be automatically redirected to the new location. This will eliminate the stress that they place on this site, and should prevent us from needing to limit the update time of the feed. We've also increased the number of items tracked in the feed from 5 to 20, so that will provide a better experience for those receiving updates.

We'll continue to monitor this new solution for issues and performance, and will make tweaks as necessary. If anyone experiences any issues, please let me know. Thanks for you feedback and reports about the RSS woes! They were really helpful :) -- Emily Diehl (talk) 20:19, 6 January 2009 (UTC)

Nice, but isn't the stress to the wiki database the same? Putting the feeds on another server should really reduce that - or was it the wiki server that had problems with so many accesses? poke | talk 20:40, 6 January 2009 (UTC)

The issues that popped up weren't db access related, so yep, this should fix it. -- Emily Diehl (talk) 17:34, 7 January 2009 (UTC)

upload bug?

Whenever i try to upload a picture, it says that its empty, even though i know its not.... File:User JonathanMartin Sig.jpgJonathan 16:21, 30 December 2008 (UTC)

Yep. Erasculio 16:22, 30 December 2008 (UTC)

See the section two above ^^ -- Wyn 23:17, 30 December 2008 (UTC)

I SAW okay? god.... have they fixed it yet >.> File:User JonathanMartin Sig.jpgJonathan 00:42, 31 December 2008 (UTC)

No... when Emily posts that they have fixed it, it will be fixed :P Patience grasshopper..... -- Wyn 00:48, 31 December 2008 (UTC)

(Reset indent) Hi guys! This issue should be fixed. If you still notice problems, let me know and I will pass the reports along. Sorry for the delay! -- Emily Diehl (talk) 19:06, 6 January 2009 (UTC)

I'm testing it right now, but it still doesn't seem to work. poke | talk 19:23, 6 January 2009 (UTC)

Not working for me either. -- Wyn/talk 19:25, 6 January 2009 (UTC)

Me neither. :S -- Indochine talk 19:30, 6 January 2009 (UTC)

Same here. Even cleared my cache before trying but it keeps failing with the EmptyFile Message. --SilentStorm 19:34, 6 January 2009 (UTC)

Hmmm, no good. Thanks for the feedback...I'm chatting with IT about this now, so I'll let you know what I find out. -- Emily Diehl (talk) 19:38, 6 January 2009 (UTC)

(Reset indent) Hi all! I wanted to give you an update on this issue. The team has been looking into it today, and hasn't been able to nail down the issue just yet. Unfortunately, there are no errors in our logs or obvious reasons why this problem has popped up. They're continuing to investigate, and hope to have the issue resolved soon. I know what a pain this particular issue is. Thanks for your patience as we try to figure out what the heck happened. I'll keep you updated and let you know when I find out more details. -- Emily Diehl (talk) 23:52, 6 January 2009 (UTC)

Well, I spoke too soon. Right as I posted that, I received an email letting me know that the issue has been tracked down and resolved. We've been able to verify that it is fixed, so please test it out and let me know if anyone experiences any additional issues. Thanks again! -- Emily Diehl (talk) 00:04, 7 January 2009 (UTC)

Emily, this problem came right after I had uploaded 100+ big (as in, really really big) images. Given how those were less than half the images I have to upload for that project...Was that the cause of the problem? Is there something we could do to avoid such thing in the future? I'm not eager to begin uploading again if it's going to break the wiki. Erasculio 00:17, 7 January 2009 (UTC)

I don't believe that had anything to do with it, although we're not sure what caused the problem exactly. From what I understand, the issue arose from a specific directory not being specified in a location. The weird thing is that this was never set before and it wasn't an issue until this point. Regardless, the issue has been corrected, so we should be good. Definitely don't let the incident affect the content you post to the site. If something breaks in a strange way, we'll just figure out what's wrong and fix it. It's part of the fun of running a live wiki I suppose :) -- Emily Diehl (talk) 01:42, 7 January 2009 (UTC)

Thanks, Emily : ) Erasculio 02:40, 7 January 2009 (UTC)

Erasculio... next time, just don't do it when ArenaNet is going to be gone for two weeks :P -- Wyn/talk 05:04, 7 January 2009 (UTC)

Special:NewImages

I keep getting a 404 when trying to access that Page. Any other Page I tried worked fine so far. --SilentStorm 00:19, 7 January 2009 (UTC)

Hiya Silent! This was an issue stemming from a problem with one of the new extensions that was installed today. It has been identified and resolved, so the problem should be taken care of. Thanks for the report, and let me know if you run into anything else that seems funky. -- Emily Diehl (talk) 01:44, 7 January 2009 (UTC)

Image problem lingering?

Maybe I'm being too overzealous, but I'm getting the internal error when uploading some images still. Displayed text as follows:
Could not rename file "public/8/83/User_Wandering_Traveler_Sensical_Measures.jpg" to "public/archive/8/83/20090107015110!User_Wandering_Traveler_Sensical_Measures.jpg"..
Any ideas? -- Wandering Traveler 01:52, 7 January 2009 (UTC)

Don't know about the other admins, but I'm having trouble deleting Image:Image-User WeHeartKatamari sig.jpg and several other images. Something about directories not being writable. -- ab.er.rant 02:13, 7 January 2009 (UTC)

Looks like it was fixed. Could have been a lingering problem. -- Wandering Traveler 05:53, 7 January 2009 (UTC)

Problem, not sure what

This is what I got when I tried to look at User talk:Emily Diehl. Image:User Shadowphoenix BUG3.jpg --Shadowphoenix 15:03, 7 January 2009 (UTC)

Hiya Shadow! Thanks for the report...this has been identified and fixed. -- Emily Diehl (talk) 17:34, 7 January 2009 (UTC)

POST limit - again

Hi Emily, again we have the problem with the POST limit that the server has (see also Guild Wars Wiki:Reporting wiki bugs/archive 8#NCsoft Error again - Important!). This makes it impossible to remove single revisions that contain too personal information on the wiki from a single page, when the page is very long. poke | talk 14:43, 30 December 2008 (UTC)

Hiya Poke! A ticket has been put in to raise the limit on the site, so I'll let you know when the changes are completed. As a side note, I thought it was funny that you actually passed the limit by one. No joke...hehe :) -- Emily Diehl (talk) 19:37, 6 January 2009 (UTC)

Haha, great :D On that topic, we added a request for a setting change, so we can remove single revisions in future; increasing the limit wouldn't be that urgent. poke | talk 19:39, 6 January 2009 (UTC)

The wiki was very slow, but it worked; thanks for the fix :) poke | talk 19:53, 8 January 2009 (UTC)

Image moving

Image moving should be enabled now, but we cannot find how it is enabled. There is no tab to move and Special:MovePage does not accept images. It gives a permission error, both here and on Gw2w. Both newly uploaded images and old images have been tested, but no luck so far. - anja 16:30, 7 January 2009 (UTC)

Hiya Anja! I talked with the team about this, and they confirmed that the proper changes were made to the site. Hypothetically, this should allow image moving. Unfortunately, even MediaWiki admits that the feature is experimental, so if it isn't working as expected, it's likely due to a MediaWiki issue. This variable just popped up in the 1.13 version, so I'm guessing that they'll stabilize it in later versions. -- Emily Diehl (talk) 17:37, 7 January 2009 (UTC)

Oh, for the archive: This was resolved btw; the right to move images was added to the group autoconfirmed, and so it underlies the same requirements of normal page moves. poke | talk 17:41, 8 January 2009 (UTC)

Bug in image uses counter

It seems that the linking pages counter at images did not work properly – it always shows 101 if there is a lot of links – example. Faalagorn☎/⚔ 00:30, 20 January 2009 (UTC)

Thats the first 101 links. (View more links to this file.) Dominator Matrix 00:32, 20 January 2009 (UTC)

And that particular image is on every single page of this wiki. -- Wyn 00:33, 20 January 2009 (UTC)

No, it's not (unfortunately). The one at the bottom of the page is called "arenanet-logo.jpg", not "Arenanet-logo-400-transparentbg.png" and it's not a wiki image ([[Image:]] ). I know it shows the first 101 (why 101 BTW?), but it should be noted that that's not the all pages. It made me confusing once. Faalagorn☎/⚔ 14:20, 20 January 2009 (UTC)

There is a text at the bottom: "View more links to this file." - and the text "The following 101 pages link to this file:" doesn't imply that that are the only pages... So no bug or error here. poke | talk 16:57, 20 January 2009 (UTC)

Guild Wars Wiki:Not orphaned

I can't seem to load this page, I am told that the server times out before the page can load. (using Firefox btw, if that helps) --Shadowphoenix 04:20, 7 February 2009 (UTC)

I get the same, could be that the list includes a number of larger images which don't fully load before the page times out like what happened with Special:NewImages a while back. --Kakarot 04:35, 7 February 2009 (UTC)

Changed the page to not show the DPL list (as that is not needed, we have the category) and to remove the gallery as that makes the same problem as it does on the NewImages special page (with creating thumbnails for too many new images). So fixed. poke | talk 18:36, 7 February 2009 (UTC)

Image cache

I hate to say it, but over this past weekend, the image cache has gotten really bad again. Some images are taking as much as 48 hrs to resolve after being updated. I don't know if this is because so many people were uploading images of the New Year parties, and related items, but it's now Wednesday, and it's not much better. Maybe take another look? -- Wyn 03:28, 5 February 2009 (UTC)

Ack :( Thanks Wyn! This has been passed along. -- Emily Diehl (talk) 20:34, 2 March 2009 (UTC)

Well, that was a month ago, and it's been much better lately, so probably not much to look into. It just got really bad after Canthan New Year for a while. -- Wyn 20:45, 2 March 2009 (UTC)

Alright, well if it pops up with a vengeance again, please let me know. -- Emily Diehl (talk) 20:55, 2 March 2009 (UTC)

Deadlocks Back

Just got a dead lock

(SQL query hidden)

from within function "SiteStatsUpdate::doUpdate". MySQL returned error "1213: Deadlock found when trying to get lock; try restarting transaction (172.30.129.98)".

When trying to browse the recent changes. Getting it several times now all over the place. Dominator Matrix 00:24, 9 February 2009 (UTC)

Hiya Dominator! I've received a response about this. Unfortunately, there's no good way to completely stop deadlocks. We've taken steps to prevent them, but it seems that the sitestatsupdate entry is an additional culprit that we think causes problems. We could disable the page counters to see if that helps, but that's up to you guys. I'm not sure if having that information is valuable to you or not. Just let me know what you think! If it's a massive problem, we can take that step. If it's an infrequent occurrence, we may just want to let it slide for the sake of the stats. Hope that helps! -- Emily Diehl (talk) 20:39, 2 March 2009 (UTC)

It's infrequent so its fine. I just reported it here so you guys could get a history. Though I'm sure you guys watch this wikis health like a hawk :). --Dominator Matrix 00:08, 4 March 2009 (UTC)

Protecting

An update for this issue. We still get the same problem when something is protected - the wiki's inaccessible (or really really slow) for all viewers. I don't think the hardware updates fixed anything - this problem happens for almost every protection. -- Brains12 \ talk 23:01, 17 January 2009 (UTC)

Thanks Brains! I'll pass this along. -- Emily Diehl (talk) 03:25, 22 January 2009 (UTC)

Heya Brains! I got a response back about this. IT isn't sure what could be causing this problem, so they'd need to have information like the time, date, and page that this happens on to check in the logs. They're guessing that the issue has something to do with a large DB query, but would need to do more investigation to know for sure. Next time this happens, could you do me a favor and let me know the page that has issues, along with the date and time of the occurrence? Sorry for the slow response on this one :( Thanks again! -- Emily Diehl (talk) 20:24, 2 March 2009 (UTC)

I protected a page today and it worked quite well, without breaking the wiki for minutes. I don't know if the problem is fully gone though (we should test it).

When I browsed through the code I got an idea that maybe could be the issue. The only area where such a loading time could come in the protection process (without preventing the page from being protected though) is the check if previous protections are run out (by the given expiration date). But as that also happens when purging a page (which works fine) I'm not sure if that really is the problem. poke | talk 20:42, 2 March 2009 (UTC)

Since it seems to be OK, I'm going to go ahead and archive this. If it pops up again, let me know and we'll dig into it more. -- Emily Diehl (talk) 19:55, 4 March 2009 (UTC)

Sysop tool: Moving a page to an existing target.

Not urgent, but still worth to report: When moving a page, as a sysop, to a target where already a page is, you get the chance to delete the target page together with the move. This works correct, but when specifiying the move target and MediaWiki detects that the target page already exists and you get the confirmation mask, the system message says "The destination page "Object" already exists. [...]" - where "Object" is a link to the article with the same name. Also the input field in that dialog for the target page is empty, so the reason for that is probably some data loss from the move page to that confirmation page.
The code that is generated when that message is on lines 111ff. in includes/specials/SpecialMovepage.php and the target title should be stored in $this->newTitle. Maybe that helps. poke | talk 23:14, 7 January 2009 (UTC)

Thanks Poke! We're checking into that for you. -- Emily Diehl (talk) 20:42, 8 January 2009 (UTC)

This might also be the problem that image moving breaks; at least moving a page and deleting the target page fails (= it deletes the target page but saying something like "file not found", it appears in the deletion log but isn't deleted). poke | talk 20:59, 8 January 2009 (UTC)

Just tried (image) moving again and it failed with "Could not find file "$1".". poke | talk 21:00, 8 January 2009 (UTC)

A recent attempt at image moving successfully moved all content on said image page...except for the image itself. In other words, it moved the tags and created a redirect at the new location, but did not move the actual image. -- Wandering Traveler 21:06, 8 January 2009 (UTC)

Sorry for my absence, guys! I'll go ahead and pass this along again for you. -- Emily Diehl (talk) 03:24, 22 January 2009 (UTC)

I've seen it move the tags, and create the redirect, but then just put a link to the image... you click on the link, and it's there, just doesn't show up on the image file page.-- Wyn 04:13, 22 January 2009 (UTC)

(Reset indent) Hi guys! I received word back about this from IT. Even though this was recently enabled on the wiki, the code is admittedly experimental...even MediaWiki says that it may not work sometimes, lol. Because of this, there's not a whole lot we can do on our end until an official fix comes out that addresses the issue. If you guys run across a patch for this, please let me know. We can test it out on the development environment to see if it works. If so, we can push it through to live. -- Emily Diehl (talk) 20:31, 2 March 2009 (UTC)

No this has actually nothing to do with image moving (which is why I talked about pages in general at the beginning). As a sysop, with deletion and move rights, you have the option, if the move-target page already exists, to check an option to delete the target page while moving. If that is the case, you get a confirmation message before the whole thing is actually performed. However when you get to that confirmation page it seems that all data that was previously sended got lost. This happens to image moves but also to normal page moves where the target page exists. But if an update to 1.14 is planned anyway, this will hopefully fix it. poke | talk 20:38, 2 March 2009 (UTC)

Ah, I see what you're talking about now. Well, let's see what happens after the upgrade. If there are still issues, we can see if we can figure out what's going on. In the meantime, I'll pass this back over to IT and have them test it out on stage to see if they can replicate the problem there. Thanks for the clarification, Poke! -- Emily Diehl (talk) 20:41, 2 March 2009 (UTC)

Heya Poke? Would you mind checking this out again to see if it's still happening? We weren't able to replicate it, but we probably didn't test the exact same things you did before. -- Emily Diehl (talk) 19:54, 4 March 2009 (UTC)

Yep, that seems fixed :) poke | talk 21:09, 4 March 2009 (UTC)

Images

It seems images are fluctuating between older and newer versions. Not only has it been affecting the recent set of skill icons I've uploaded but images like Image:Everlasting Abyssal Tonic.png and Image:Warrior Monument armor f gray front arms legs.jpg as well. Sounds Risky | 07:30, 7 March 2009 (UTC)

lol? When I'm on File:Warrior Monument armor f gray front arms legs.jpg and press purge the size of the image, Wyn uploaded, changes every time o.O But I don't get the actual cropped image. poke | talk 14:44, 7 March 2009 (UTC)

Is there an actual difference in the images, because for me both images are identical including pixel size but Wyn's is 117kb whereas Raine's is 39kb? --Kakarot 15:20, 7 March 2009 (UTC)

The image cache has been wacked again every since the upgrade. I'm thinking it might be all the extra thumbnails that 1.14 introduced on each file page. And yes Kakarot, the images are vastly different. -- Wyn 15:24, 7 March 2009 (UTC)

File:Warrior_Monument_armor_f_gray_back_chest_feet.jpg Even trying to revert is wierd, look at the sizes etc. -- Wyn 15:28, 7 March 2009 (UTC)

(Edit conflict) ... For a second I was getting what Poke had, so I came back here to mention it but before saving I checked the image and it was back to identical sizes. But yeah, the new thumbnails added in 1.14 could be the cause as it seems to have happened not long after the upgrade. --Kakarot 15:30, 7 March 2009 (UTC)

Ok, since the upgrade, the image cache is once again a mess. Poke tells me it's not the thumbnails causing the problems so I don't have any other ideas. Images aren't refreshing when they are updated. Gallery_of_female_necromancer_Krytan_armor and Gallery of female warrior Monument armor specifically.... it does weird stuff, sometimes just applies the new dimensions to the old image, sometimes applies the old dimensions to the new image, sometimes bounces back and forth. -- Wyn 20:43, 8 March 2009 (UTC)

The last set of skill icons I uploaded was about a day before the software upgrade and I had actually noticed the fluctuating images then, before the upgrade. I imagine the upgrade didn't help, but it seemed to be happening at least a day before. Sounds Risky | 11:08, 9 March 2009 (UTC)

Ehehehehe! We have a gallery of dumpy dwarves and lanky giants :D --snograt 14:41, 9 March 2009 (UTC)

(Reset indent) This fluctuating is still occuring. Sounds Risky | 05:02, 16 March 2009 (UTC)

New version uploaded at the beginning of the week: Ahtok.jpg has an updated thumbnail (alive), but full-sized picture stays the same (ghost; same as Scout Ahtok.jpg). --NIN37 07:46, 20 March 2009 (UTC)

Blah! Much hate for image caching issues :( I passed this stuff along when Wyn first messaged me on my page a while back, and sent it along again tonight. Are you guys still seeing this happen a lot, or are the occurrences scattered? I haven't had time to do mass uploads lately, so I haven't been able to see it first hand this time. -- Emily Diehl (talk) 05:08, 25 March 2009 (UTC)

Almost everyone that uploads new versions of any image has to be told to stop reuploading or reverting because they just don't change. I find that if you just upload and leave it, it will resolve itself most of the time, but some of those armor images I waited hours and now weeks and they still haven't fixed themselves. -- Wyn 05:10, 25 March 2009 (UTC)

File:Ahtok.jpg still displays wrong here: this time, thumbnail [1] & full-sized [2] are incorrect (blue npc), but if I click on thumbnail and see picture itself [3], it shows right (colorful npc). Yeah, I know there are other examples, but I think this one is easy to spot since colors are really different :P

File:Scout Ahtok.jpg does not (seem to) have this issue, because both full-sized picture and real picture are 225 pixel wide ("No higher resolution available"). Not sure if this helps (in other words, seems the "thumbnail generator" is "miss-behaving"). --NIN37 06:52, 25 March 2009 (UTC)

(Reset indent) Heya guys! We've made some more adjustments to our caching settings, so please let me know if this still pops up. Thanks :) -- Emily Diehl (talk) 23:51, 25 March 2009 (UTC)

Heya again all! I just wanted to pop in and give you a heads up about something. So the issue that caused the wiki to burp for a while yesterday pointed the team at another larger issue pertaining to a cache corruption, which they believe was also what was causing this recent batch of image problems. The good news is that they've resolved that issue, so we're hoping that this takes care of the image problems as well. We're going to keep monitoring, so let me know if anything pops up again so I can pass the report onto them. Thanks again, and sorry for the hassle! -- Emily Diehl (talk) 00:08, 27 March 2009 (UTC)

How do you suggest we correct those image that are still "stuck"? Reupload them? or delete/rename? -- Wyn 00:27, 27 March 2009 (UTC)

If you guys are able to do that, I'd say that's the way to go. You could also send a list of them to me and I can pass it along, but the team wouldn't be able to get to deleting them immediately. -- Emily Diehl (talk) 00:52, 27 March 2009 (UTC)