For about the last three days, lowercase sigmabot III has only been archiving the Administrator's noticeboards and nothing else. Somebody mentioned that you gave it a good kick the last time it went on the fritz, so I will go ahead and notify you. Safiel (talk) 16:37, 29 April 2024 (UTC)[reply]
HiAssem Khidhr, my apologies for not replying to this sooner, but as you probably guessed by my lack of response I don't have the free time to work on this task at the moment. Sorry. — The Earwig (talk) 04:23, 7 June 2024 (UTC)[reply]
(Sorry if this is the wrong forum for asking, but if so, perhaps you could point me in the right direction?)
I use the Copyvio Detector (great tool, BTW!) in checking new AfC drafts, at least a dozen times most days. I sometimes get an error message saying that the detector has exceeded its maximum allowed Google searches. This issue has always been there, occasionally, but in the last week or two it has occurred daily. When I start reviewing, around 6am or so UK time, the first few reviews always hit this problem. Then, maybe 8am (?) the daily quota probably gets reset, or something else happens, because from then onwards everything is fine until the next morning.
So I was thinking, I don't suppose there's much we can do to increase the quota (?), but would it be possible to add another search engine as a fallback option? Either so that when the user gets that error message, they could manually tick a box to use Bing (say) instead; or maybe the Detector could automatically switch to using the alternative if Google has failed.
HiDoubleGrazing, using Bing or some other engine as a fallback is definitely something we’ve discussed—I hadn’t realized the issue had gotten this bad recently. The main issue here is these services usually cost money, and while the WMF pays for our Google access right now, I don’t know if I will be able to ask for access to additional search engines. First, I can take a deeper look into whether anyone is overusing their share of the tool’s resources; we might need to block/limit them. (Our plan with Google allows about 1500 articles to be checked per day.) — The Earwig alt (talk) 16:11, 8 May 2024 (UTC)[reply]
Okay, thanks for shedding some more light on this; needless to say, I knew nothing about how these things work.
I guess we at AfC are taking up quite a chunk of that quota, given that we see what are by definition new drafts usually by new users. I for one run the check probably at least on ⅓ of the drafts I review (and if you think that makes me an overuser, feel absolutely free to point this out, of course!). Even at NPP we deal with relatively more experienced users, so there's that much less of a need to check for CV.
It may be that I see the problem worse than some others, mind, because of my weird early-morning AfC habit, combined with the time zone I'm in. -- DoubleGrazing (talk) 17:05, 8 May 2024 (UTC)[reply]
Hi again,
Quick update on this, the problem (of the copyvio detector running out of Google quota) has lately become worse. Unlike before, when it would only manifest in the early morning UK time, and usually be fine after 8am UK / 0700 UTC, it's now happening also in the afternoon. This is relatively new, maybe in the past week or two, so I've not yet have a good feel for what time it happens exactly (in case that matters); I would have said late afternoon, but eg. today it started already around 1pm UK / 1200 UTC.
Hello Ben, sorry to bother you so early and on a Sunday. The Copyvio detector seems unable to perform any comparisons at the moment. It sits and spins for three minutes before timing out ("The URL https://www.bbc.com/news/articles/cz55y6k0p5go timed out before any data could be retrieved.") Any assistance appreciated, as we have a lot of reports at CopyPatrol, a lot more than usual, and we will not be able to assess them without this tool. Thank you! — Diannaa (talk) 11:48, 2 June 2024 (UTC)[reply]
@The Earwig: It's down again as of 6 June 2024. It takes a long time to reach and then after entering the page title and clicking submit in runs after several minutes with 0 errors. I've tried this with other articles, that got higher vilolations before. Thanks for any help you can provide. Greg Henderson (talk)09:06, 2 June 2024 (UTC)[reply]
Today, getting the error message: "An error occurred while using the search engine (Google Error: HTTP Error 429: Too Many Requests). Note: there is a daily limit on the number of search queries the tool is allowed to make. You may repeat the check without using the search engine." Greg Henderson (talk) 23:14, 7 June 2024 (UTC)[reply]
(talk page watcher) @Greghenderson2006: This happens when we've reached our daily quota with Google. Unfortunately, the copyvio detector can only handle up to around 1,250 a day. You'll need to try again after a few hours or so. In the meantime, you can try using the copyvio detector without search engine checks, which will still work. Chlod (say hi!) 01:07, 8 June 2024 (UTC)[reply]
I have a question about this editing tool. It seemed like I could run this 20 or more times before I got a notice that I had reached my daily limit. But now, I receive a notice if I just run it a few times. Has this limit been decreased for some reason? I use this tool quite a lot while patrolling drafts and CSD categories so it's sometimes difficult to remember to go back to reexamine some pages the next day when I have reached my daily limit for the current day. Thanks for any insight you can provide. LizRead!Talk!20:21, 8 June 2024 (UTC)[reply]
HiLiz. Rest assured this isn't related to your own usage of the tool. The daily limit is shared by all users, and allows for about 1000–2000 pages to be checked per day, so even if you're checking a few dozen, that's not a major contributor to the limit getting reached. We've been noticing this issue more frequently recently (see a few threads above) and we're doing some work to restrict other users of the tool who are actually overusing their share of its resources. I'm hoping to have things back to normal soon. — The Earwig (talk) 04:23, 11 June 2024 (UTC)[reply]
I've started using the Arc browser, for some reason whenever I try and access Copyvios on it, I get an Internal Server Error. Trying the same URL in Edge works fine. Not sure where the bug is there, but hopefully you can find it.
Hey Rich, sorry I took a bit to reply. This is my first time hearing about Arc and I don't really feel like creating an account to test, so I can't confirm on my end. Are you sure it's an Internal Server Error or may it be a 403 Forbidden? (We may have inadvertently blocked its user agent as a crawler, which would give a 403, but I don't see anything in our block list that looks like it or Chrome [except Linux], so I don't know.) This is pretty strange.
Regarding bot usage, there are two main issues the tool's had lately: general downtime and exhausting our Google credits. I've improved the tool's performance a bit so the former is not a major issue now, but we are still frequently exhausting our daily Google quota. I've checked RichBot's usage and recently it's been consuming around 10-20% of our total Google credits. That's not too excessive, but if you could find a way to tone it down a bit compromising its usefulness, it would be appreciated. — The Earwig (talk) 08:10, 1 July 2024 (UTC)[reply]
No worries, I have reduced RichBot to only look at 100 (plus existing CVs) per run, so 200 per day (excluding manual runs). Is there a way we can increase the credits? I don't mind throwing some £ at it if need be - RichT|C|E-Mail09:31, 1 July 2024 (UTC)[reply]
No way that I know of unfortunately; the WMF pays for it, but Google's API terms limit our usage without some kind of special arrangement that I have been unable to get. — The Earwig (talk) 15:25, 1 July 2024 (UTC)[reply]
HiFirefangledfeathers, thank you. I'm not sure who we could talk to about this, to be honest. My former contact at the WMF no longer works there and it's not clear to me who is responsible for managing the relationship with Google right now. Going the other way, i.e. getting someone in a position of power at Google who could help, might be more fruitful. But that is just speculation; I don't know who specifically that might be. — The Earwig (talk) 06:02, 19 July 2024 (UTC)[reply]
And it's definitely a 500, 'The server encountered an internal error and was unable to complete your request. Either the server is overloaded or there is an error in the application.' - RichT|C|E-Mail14:07, 1 July 2024 (UTC)[reply]
Local administrators can now add new links to the bottom of the site Tools menu without using JavaScript. Documentation is available on MediaWiki. (T6086)
I regularly used this tool you created, mostly when patrolling drafts or CSD-tagged articles, I'd probably used it 3 or 4 times a day. When I used it too much, I'd get a message that I was over my limit of how often I could use it. At least that's how I thought things worked. Now, I get this message every time I try to see whether a page is a copyright violation, I have not gotten a successful response to a query in many, many weeks now. So, I'm wondering is this "limit" actually for all users on this platform and not tied to individual editors? Because something odd is going on and maybe new page patrollers or AFC reviewers are using it for every article they review if I can not just get one or two reports on suspicious articles or drafts I've come across. I know with AI, there are ways users can get around copyright restrictions but I still found the tool helpful.
Do you have any idea why it is suddenly no longer available to generate reports? Can you tell me the time of the day when it "resets" so that maybe I could make inquries then? Or is there any possibility of raising this limit of reports generated? I mean, I'm glad it's become so popular but it has also become unavailable for use for those of us who just want to make a few queries a day. Thank you. LizRead!Talk!22:31, 19 July 2024 (UTC)[reply]
HiLiz, truly sorry about the ongoing issues. I'm aware and working on it (see some of the threads above you), with the time I have available. I thought things has improved with the overall performance improvement last month, but it has really just made this particular problem of running out of the search quota much worse. Anyway, I am working on it now.
To answer your questions: yes the quota is shared by all users, and we cannot easily raise it. It's a hard limit enforced by Google that I cannot bypass without some special arrangement. It resets I think around midnight Pacific Time, i.e. Google's time zone.
I think the issue is some bots/automated traffic making too many queries. In the past I have been able to block them or ask them to slow down, but that approach has become less effective lately. So, I will be adding authentication to the tool to make sure only logged in users can use it and I can more accurately identify who is overusing it. I expect to finish that work this weekend and I am hopeful that will solve the issue. If it doesn't, there are other things I can try. — The Earwig (talk) 00:43, 20 July 2024 (UTC)[reply]