The Signpost

Technology report

Efforts to "normalise" Toolserver relations stepped up

Contribute  —  
Share this
By Jarry1250

Toolserver parties try to inject festive goodwill into dispute

I'm looking forward to working with Wikimedia Germany, other Toolserver supporters, DaB, and the tools and bots communities to help improve their experience.

—WMF Engineering Community Manager Sumana Harihareswara this week

Efforts were stepped up this week to sow a feeling of trust between the major parties with an interest in the future of the Toolserver. The tool- and bot-hosting server – more accurately servers – are currently operated by German chapter, Wikimedia Deutschland (WMDE), with assistance from the Wikimedia Foundation and numerous volunteers, including long-time system administrator Daniel Baur (more commonly known by his pseudonym DaB). However, those parties have more recently failed to see eye-to-eye on the trajectory for the Toolserver, which is scheduled to be replaced by Wikimedia Labs in late 2013 (see Signpost coverage: 1, 2), with increasing concern about the tone of discussions.

At the crux of the disagreement is a single question: whether Wikimedia Labs can ever viably supplant the Toolserver. The Foundation is certainly throwing its weight behind the transition, but it is people like DaB – not to mention hundreds of Toolserver volunteer developers – that will need to be convinced if content is to be properly migrated, given the substantial switching costs individual developers must bear. The switching process will be further complicated if those developers share the concerns of DaB about the long-term prospects for Labs. It is not an easy situation to manage, especially given the tight timetable.

Two positive steps forward were taken this week. In a meeting on Thursday, WMDE community liaison Denis Barthel and DaB agreed that the WMF plan should be taken in good faith as a viable proposal for transition, and deserves the investment of time and energy. Though DaB later shared his ongoing doubts on the Toolserver-l mailing list, he offered his support for users wishing to transition, though he said he could not advise on setting up tools on Wikimedia Labs; for that, developers would have to rely on WMF help.

24 hours later more news reached the ears of Toolserver developers: it was announced that that WMF help would be co-ordinated by WMF engineering community manager Sumana Harihareswara. Harihareswara is one of the Foundation's most recognisable faces, regularly attending developer meetups both in person and online. As a senior WMF employee, her appointment to the role is likely to help resolve a feeling that the feelings of individual Toolserver developers – who may not appreciate the powerful range of environmental options available on Labs – were being ignored by WMF senior management. This move can only help ease tensions between the parties.

Wikimedia Germany, for its part, remains essentially stuck in the middle, committed both to keeping the Toolserver going in some circumstances and shutting it down in others. It can be little wonder the organisation is seeking to "normalize the relationship between the Toolserver, WMDE and the Wikimedia Foundation". Fortunately for the chapter, that middle ground may have just become a slightly nicer place to be.

In brief

Not all fixes may have gone live to WMF sites at the time of writing; some may not be scheduled to go live for several weeks.

+ Add a comment

Discuss this story

These comments are automatically transcluded from this article's talk page. To follow comments, add the page to your watchlist. If your comment has not appeared here, you can try purging the cache.

That is good news in this Holiday season! Everytime I try to understand the whole Toolserver problem I keep thinking of that line from the movie Cool Hand Luke where the guard beats up a prison inmate and then calmly explains "What we've got here is (a) failure to communicate". Jane (talk) 10:30, 26 December 2012 (UTC)[reply]

No offense, and I'm not sure what engvar applies to the Signpost but it's normalise, quoted, in the header, and normalize, in an unattributed quote, in the body. Perhaps it would be best to avoid both usages.--Wehwalt (talk) 11:02, 26 December 2012 (UTC)[reply]

Wehwalt, I'm pretty sure we're not permitted to change spelling variety in quotes (MOS). It does lead to clashes within text, but there's no perfect way. Tony (talk) 13:15, 26 December 2012 (UTC)[reply]
But my point (badly expressed, I'm afraid) is that you did just that. Which is the quoted word? Normalize or normalise?--Wehwalt (talk) 14:04, 26 December 2012 (UTC)[reply]
Quite. I'm afraid I was sufficiently ambivalent about changing the original z to an s (all Tech Reports I write are in British English for personal convenience) that I changed my mind halfway through. Correct now though :) Happy tidings, - Jarry1250 [Deliberation needed] 16:02, 26 December 2012 (UTC)[reply]
  • The way I look at it, as interested person but not a developer of any kind, was that I was rather a bomb that was dropped, then a bit of negotiation about the end of the Toolserver. But I am aware I don't know all the ins and outs. Main question is, why the tight schedule of conversion? That Wikilabs will be live soon, okay. But why immediately closedown Toolserver? Can the transition period not a bit longer (an extra year or so) to give less pressure and stress? When tools and bot are gradually moved to Wikilabs, the load on Toolserver will also gradually get lower. Most likely, that will make investment in new equipment superfluous, reducing costs for Toolserver. The way I look at it, Sumana Harihareswara has a nice challenge on his hands. First, she will have to restore confidence between the Toolserver lads and lasses. Second, she will need to start a kind of Outreach and or Helpdesk, to assist the tool and bot managers to move over. I have every confidence that she will be able to do that, but not in the present, limited time schedule. The Banner talk 11:11, 26 December 2012 (UTC)[reply]

If the production database replication and access (i.e., cross-database joins and joins with user databases) turns out to work better, faster, more reliably, or with less lag on Labs than the Toolserver, I suspect volunteers will quickly flock to it in droves. But that remains to be seen. We should know in the next three months. 70.59.14.20 (talk) 13:03, 26 December 2012 (UTC)[reply]

  • I am a Toolserver developer who runs a handful of tools from the system. I'm used to the way it works and it's fairly simple to get things done. However, with the downtime in 2012 and the prospect of it being gone, I thought I would migrate to Labs. I talked with the Labs team and set up an account and a million other things in a complex system of projects, instances, permissions, keys - at the end of the day the idea of running tools on a LAMP stack was really a lot of work. I'm still at the Toolserver. Mono 18:28, 26 December 2012 (UTC)[reply]
Thank you, Jarry1250 - Nabla (talk) 05:01, 27 December 2012 (UTC)[reply]
  • I have always been a bit worried that Toolserver was a risk point for the project. I am however more worried that WMF is trying to Borg it. Freedom to fork is critical, having Toolserver independent helps that freedom. I am also concerned with the bureaucratic overhead Mono mentions above. Rich Farmbrough, 01:21, 27 December 2012 (UTC).[reply]




       

The Signpost · written by many · served by Sinepost V0.9 · 🄯 CC-BY-SA 4.0