Jump to content


Rendered with Parsoid
From Wikitech
This talk page is not a Cloud VPS or Toolforge support forum. Contact information for Cloud VPS and Toolforge can be found at Help:Cloud Services communication.

Where to a go for user help?

Have been using the CatScan tool. Last day or so the results are out of date or simple not correct (or at least incomplete). Who do I ask about the status of the database update? --Traveler100 (talk) 07:09, 19 November 2016 (UTC)Reply

Which CatScan tool are you using? Generally the maintainer list for given tool can be found at https://tools.wmflabs.org/?tool=NAME_OF_TOOL where NAME_OF_TOOL is taken from the URL of the tool that you are interested in (e.g. https://tools.wmflabs.org/?tool=sal to find the maintainers of https://tools.wmflabs.org/sal/). --BryanDavis (talk) 22:20, 19 November 2016 (UTC)Reply

tools login

There is no login link at https://tools.wmflabs.org/. So if you've got a login and have already set up an SSH key but you're not logged in to toolsadmin, there's no obvious way to create a new tool - clicking the 'Create a new tool' link results in HTTP 403. It'd be nice if either there was a login link, or /tools/create redirected to a login screen if you're not logged in. GoldenRing (talk) 08:11, 4 October 2017 (UTC)Reply

Thanks for the bug report. I have created phab:T177407 to track this problem. --BryanDavis (talk) 15:21, 4 October 2017 (UTC)Reply

Possible typo

I was led to this page from https://tools.wmflabs.org/dispenser/view/Peer_reviewer#page:Fourth_Amendment_to_the_United_States_Constitution; the page displays a possible grammatical error, highlighted in bold below:

Wikimedia requires notification of when leaving the Foundation's sphere of influence.

The word of seems unnecessary here.--NibKing (talk) 04:48, 6 September 2018 (UTC)Reply

@Dispenser: FYI ^. Quiddity (talk) 18:21, 6 September 2018 (UTC)Reply

Planned changes to organization of Toolforge Portal

There are some planned changes for the Toolforge Portal. Some elements of this page may change as information is added and reorganized. Preceding unsigned comment added by SRodlund (talkcontribs) 01:02, 14 May 2019 {UTC)

unable to login

I am currently unable to login using putty, as well as winscp from one ISP. But I am able to login using other ISP without any issues. The problematic ISP's IP range is blocked on enwiki on vandalism, but not mrwiki. Any idea why is this happening? I login on dev-buster.toolforge.org —usernamekiran (talk) 16:36, 8 September 2022 (UTC)Reply

Addendum: If I visit my tool's web-service link I am getting "The connection has timed out" error from problematic ISP. From the good ISP, the link/page is opening properly: "No webservice. The URL you have requested, https://kiranbot4.toolforge.org/, is not currently serviced." —usernamekiran (talk) 17:10, 8 September 2022 (UTC)Reply

IP range blocks on wikis should not have any effect on connecting to Toolforge bastions or hosted tools.
Both ssh and https not working from the same ISP, but things working from another ISP you have access to sounds like an ISP specific connectivity problem. There are instructions on Reporting a connectivity issue which might help you gather enough information for a Network engineer to figure out what is going on. -- BryanDavis (talk) 17:19, 8 September 2022 (UTC)Reply
Thanks Bryan. I will look into it. I just realised that xtools as well, is not loading from the problematic ISP. —usernamekiran (talk) 14:09, 12 September 2022 (UTC)Reply

Availability stats for geo services (osm4wiki, wikimap, geohack)

Hi BryanDavis, I am addressing you because you seem to be able answering most queries here. On de:WP, we are having a discussion about how to best display information about works of art in the public space. The discussion started because some members of the volunteer team working on this wanted to provide a back-up in case neither of the geo services osm4wiki, wikimap, geohack were working. I have noticed problems with these services every now and then, but I am not sure whether they are specifics to the service or the toolforge platform in general. Lately, it seems to have gotten better, but my sampling rate is too low. Could you provide us with availability stats for these geo services? I am not asking about SLAs in some form or contract, just actual data on uptime (in %), preferably broken down per month. Thank you very much! --Minderbinder (talk) 10:42, 30 July 2023 (UTC)Reply

@Minderbinder, unfortunately neither Toolforge nor Cloud VPS directly track availability of individual webservices. Toolforge tools that are using the Kubernetes platform have some operational statistics collected at https://grafana.wmcloud.org/d/TJuKfnt4z/kubernetes-namespace. You could also try to make some queries of https://toolviews.toolforge.org/api/ to form an approximation of uptime by looking for trends in HTTP 2xx responses by a given tool over time. -- BryanDavis (talk) 17:12, 30 July 2023 (UTC)Reply
Hi Bryan, thank you for your answer. I don’t think the Grafana stats are specific enough, or maybe I am too ignorant to extract total uptime for the whole platform. The queries are nice, do they differentiate between different HHTTP response codes? I could not find that, just hit numbers and unique visitor numbers. Am I missing something? —Minderbinder (talk) 07:40, 31 July 2023 (UTC)Reply
@Minderbinder Toolviews only counts HTTP 2xx status responses from the front proxy access logs. The graphana dashboard does provide numbers on other status codes (an example from the ifttt tool I currently maintain).
There really is not anything to look at that would exactly track "uptime" generically across tools. In the Kubernetes data displayed by Grafana there are some broad signals such as the 'Pods' panel that shows how many Kubernetes Pods (groups of running Docker containers) are active in the tool's namespace, the 'Object counts' panel that shows various non-Pod Kubernetes objects, and the HTTP status code panel I mentioned previously. Depending on the particular tool these might give a sense of how it is functioning. We do not typically have Pods instrumented in a way that Kubernetes will notice when the application code is malfunctioning but still running (such as returning HTTP 500 responses or a bot that is stuck waiting on I/O for extended periods) so the signals we do track may not tell the whole story.
You may have more luck talking to the maintainers of the individual tools you are interested in directly. If nothing else that may help you determine how responsive the maintainers are to community requests. That response rate can sometimes be seen as a proxy for how responsive they are to runtime problems with their software. You can use the search box on https://toolsadmin.wikimedia.org/tools/ to find a tool and then click through to see its maintainers. Then you can use https://contact.toolforge.org/ to find ways to contact those maintainers. You can also try email contact by mailing $toolname.maintainers@toolforge.org. See Help:Toolforge/Email#Mail_to_a_Tool for more details. -- BryanDavis (talk) 20:53, 1 August 2023 (UTC)Reply

Proposing doc changes for this page and for Toolforge navigation menu

Hello, as part of a technical documentation project to improve the overall navigation of Cloud Services docs, I am proposing some minor changes to this page, along with a new navigation menu for the Toolforge user and admin docs. For details and how to provide your feedback on my proposed revisions, see https://phabricator.wikimedia.org/T341364. Thanks! Triciaburmeister (talk) 20:54, 17 August 2023 (UTC)Reply

More specific Phab task specifically for this page: https://phabricator.wikimedia.org/T344631 Triciaburmeister (talk) 17:04, 21 August 2023 (UTC)Reply