Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Provide a sandbox instance on smw.org #729

Closed
mwjames opened this issue Jan 20, 2015 · 40 comments
Closed

Provide a sandbox instance on smw.org #729

mwjames opened this issue Jan 20, 2015 · 40 comments
Assignees
@kghbln
Labels
infrastructure

Comments

@mwjames
Copy link
Contributor

mwjames commented Jan 20, 2015

Following-up on [0], it should be rather simple (given the Composer install) to provide a https://sandbox.semantic-mediawiki.org/ instance for users to play and test without much cost (from a technical and human effort perspective).

@kghbln For spam protection maybe Karsten can weigh in on what needs to be done.

I prepared a XML that can be used (via Special:Import or importDump.php) to import and setup some examples and/or for resetting the sandbox to an initial state.

[0] http://wikimedia.7.x6.nabble.com/Defining-duplicate-property-names-in-subobjects-tp5042657p5042674.html
@JeroenDeDauw
Copy link
Member

This needs to be discussed on the OSDA list. @mwjames can you send emails to this thing? Perhaps not if you are not on it. In can always send one and cc you.

@mwjames
Copy link
Contributor Author

mwjames commented Jan 20, 2015

This needs to be discussed on the OSDA list.

Not on the list.

@mwjames can you send emails to this thing? Perhaps not if you are not on it. In can always send one and cc you.

I hate administration, if someone can link to this issue I would be much obliged.

@JeroenDeDauw
Copy link
Member

Apparently https://sandbox.semantic-mediawiki.org/ existed in the past. Not sure what happened to it. Ping @kghbln @mkroetzsch

@JeroenDeDauw
Copy link
Member

I asked @yaronkoren if referata (and thus http://scratchpad.referata.com/) would likely be updated in the next month or two, and he said probably not.

@kghbln
Copy link
Member

kghbln commented Jan 22, 2015

https://sandbox.semantic-mediawiki.org/ was abandoned since it was unmaintained and spammed. At that time we still lived with the pre Composer 1.8 world and figured that http://scratchpad.referata.com/ is the better and far more popular alternative. One reason for sandbox.smw.o not being maintained was that it was still "handled" via KIT and not OSDA. Now it redirects to semweb.o. which was not really the best choice I guess.

Basically what is needed is the one in control of the DNS for sandbox point it to the OSDA server and do a MW install + the Composer stuff for SMW. @mkroetzsch has the best insight here I suppose.

@JeroenDeDauw
Copy link
Member

redirects to semweb.o. which was not really the best choice I guess.

That wiki is down?!

the one in control of the DNS for sandbox point it to the OSDA server

Is it not OSDA that owns the domain? In that case we ought to have DNS control over all subdomains no?

@kghbln
Copy link
Member

kghbln commented Jan 23, 2015

That wiki is down?!

Not a couple of day ago.

Is it not OSDA that owns the domain? In that case we ought to have DNS control over all subdomains no?

This was my way of adding a ray of hope to the discussion that it probably will not take too long to get this act together.

@jongfeli
Copy link
Contributor

@kghbln if you need help to setup a sandbox wiki from scratch I am willing
to help out. This is what a lot of people are missing and can be of great
help for debugging SMW.
Op 23 jan. 2015 10:23 schreef "Karsten Hoffmeyer" <notifications@github.com

:

That wiki is down?!

Not a couple of day ago.

Is it not OSDA that owns the domain? In that case we ought to have DNS
control over all subdomains no?

This was my way of adding a ray of hope to the discussion that it probably
will not take too long to get this act together.

—
Reply to this email directly or view it on GitHub
#729 (comment)
.

@kghbln
Copy link
Member

kghbln commented Jan 23, 2015

@jongfeli Thank you for offering you help. That's great. If there are on-wiki things to do you may definitively do so. As a matter of fact you are very welcome to do so. Installing the wiki as such requires access to the server so @JeroenDeDauw will be our man. :) Once this is done we see that @mwjames xml gets into the wiki.

Besides, after a lot of reluctance I started upgrading my farm from 1.8 to 2.1 today and this was much less painful than I originally expected as a matter of fact no pain at all was involved so far. So I believe I currently offer the only public SMW 2.1 test wiki as of today. "Sadly" it is a German language instance.

@mwjames Is there a reason why we recommend people killing their semantic store prior to upgrading from 1.8? I do not really see a reason to do so and I have not done so without probs.

@mwjames
Copy link
Contributor Author

mwjames commented Jan 24, 2015

@mwjames Is there a reason why we recommend people killing their semantic store prior to upgrading from 1.8? I do not really see a reason to do so and I have not done so without probs.

Did we, if so I can't recall the reason.

@mkroetzsch
Copy link
Contributor

I have now deleted the subdomain DNS entry for sandbox.semantic-mediawiki.org, which means it will point to out main server (after DNS has updated everywhere). If somebody wanted to configure a sandbox wiki there, it would be possible now. I can also point to any other server if we had a sandbox elsewhere, of course.

@mwjames
Copy link
Contributor Author

mwjames commented Sep 21, 2015

@JeroenDeDauw @kghbln It would be really neat if this could be resolved by the time of SMWCon (or during).

@kghbln
Copy link
Member

kghbln commented Sep 21, 2015

I wonder why you got this idea today. ;) pass-throug ping to @JeroenDeDauw

@JeroenDeDauw
Copy link
Member

While I like to see such a sandbox happen, I think it is a bad idea for me to take up this task. Installing and maintaining wikis is not something I like doing or am particularly good at. Just look at how frequently the main SMW wiki is getting updated... Perhaps someone who has a decent server wants to volunteer to run a sandbox wiki, and then we can point the DNS for the sandbox subdomain there?

@kghbln
Copy link
Member

kghbln commented Sep 25, 2015

I really would not like to have access to the osda-server, however I was already thinking of providing a test instance on my private test server. Dunno about the load I will be getting, but we could at least test if this is a way to go.

@JeroenDeDauw
Copy link
Member

We could ask around on the mailing list. You are already doing a pile of stuff in other areas

@kghbln
Copy link
Member

kghbln commented Sep 25, 2015

Ah, I was more thinking about the server load from the traffic. I have 1 GB of RAM and half a CPU (MHz: 2266.746, L1d cache: 32K, L1i cache: 32K, L2 cache: 4096K)

@mwjames
Copy link
Contributor Author

mwjames commented Sep 25, 2015

We could ask around on the mailing list. You are already doing a pile of stuff in other areas

This is going into the wrong direction. If we cannot make a sandbox then we just let people use smw.org to test things instead of pushing them off or link them to an outdated release.

People want to try things out before they start to install the software, so let them.

@kghbln
Copy link
Member

kghbln commented Sep 25, 2015

My idea or better understanding of this sandbox is that it always runs latest stable (not master) of MW as well as SMW and (semantic) extensions. Apart from that I'd rather like more active testers than somebody agreeing to provide a sandbox.

To cut it short. I will try to set it up this weekend so that the DNS may be changed to the respective IP.

@kghbln kghbln self-assigned this Sep 25, 2015
@JeroenDeDauw
Copy link
Member

@mwjames I think you do not understand what I was getting at. My suggestion is not to point people to some random badly maintained wiki out there where they can test stuff. Rather, I'd like to see a test wiki that is actively maintained, so it does not get out of date, and so that new things that people want to test can be enabled in timely manner. Sure we can do this ourselves, though that takes our time away from other SMW things we are already doing. Having a new dedicated volunteer to take care of this would be great no?

@mwjames
Copy link
Contributor Author

mwjames commented Sep 26, 2015

Having a new dedicated volunteer to take care of this would be great no?

It would be great in an ideal world and now and then I'm surprised by a light of hope that shines through the cracks (as with #1165) but I have come to terms with the fact that nothing is happing if you are not pushing for it, contribution is limited to the existing candidates which you can count fairly easily.

While I would welcome more volunteers and have users scream "I will do it, I'll help you ...", the hard reality shows its ugly face (and that is what experience dictates not the mere hope for an utopia).

@kghbln kghbln mentioned this issue Sep 26, 2015
Closed
@kghbln
Copy link
Member

kghbln commented Sep 26, 2015

So this is done now. Just add IP 217.197.83.171 to the A-Record of "sandbox.sematic-mediawiki.org".

@mwjames
Copy link
Contributor Author

mwjames commented Sep 26, 2015

So this is done now.

Well done.

image

PS: The super contributor badge was not available so I took the next best thing.

@mwjames
Copy link
Contributor Author

mwjames commented Sep 26, 2015

@kghbln Did I miss something?

image

@kghbln
Copy link
Member

kghbln commented Sep 26, 2015

This is the default page of the testserver. First the DNS configuration has to be updated for it to work. A little patience is needed until @JeroenDeDauw worked on this.

@kghbln
Copy link
Member

kghbln commented Sep 26, 2015

Well done.

Cooool. :D

Currently "sandbox.semantic-mediawiki.org" redirects to "semantic-mediawiki.org"

@mkroetzsch
Copy link
Contributor

Great, many thanks! I have now changed the A-entry for sandbox.semantic-mediawiki.org. Works for me, but it might still take a moment until all DNS serves have it.

@JeroenDeDauw
Copy link
Member

\o/ Thanks @kghbln, @mwjames and @mkroetzsch

@JeroenDeDauw
Copy link
Member

It would be great in an ideal world and now and then I'm surprised by a light of hope that shines through the cracks (as with #1165) but I have come to terms with the fact that nothing is happing if you are not pushing for it, contribution is limited to the existing candidates which you can count fairly easily.

While I would welcome more volunteers and have users scream "I will do it, I'll help you ...", the hard reality shows its ugly face (and that is what experience dictates not the mere hope for an utopia).

Reality is indeed quite different than a stampede of people that dedicate all their free time to help out without first being further incentivized. Most people will not contribute back. Some will, though most of these need a nudge to get started, or at the very least some easy thing to start with. If we do not clearly communicate contributions of various types are welcome (in a way understandable to the people that are potentially interested in contributing), and we do not have easy tasks for those people, then indeed, it is silly to expect new contributors to appear. This is how I got involved via GSoC. It's how @jongfeli got started with submitting CS improvements. It's how several people got into the testers group and are now helping out with that. I think you'll find there are a multiple awesome people like @kghbln in our community that'd be willing to set up a wiki such as discussed here if you ask.

@kghbln
Copy link
Member

kghbln commented Sep 28, 2015

@JeroenDeDauw @mwjames
I dunno where I read it but someone suggested the sandbox to run on master when it comes to the semantic extensions located at GitHub. I can do this since it is just a matter of changing the "composer.local.json" file. I have create a cronjob to do the rest every night anyways.

@kghbln
Copy link
Member

kghbln commented Sep 28, 2015

This was on twitter by @JeroenDeDauw. Since @mwjames also indicated via e-mail that master would be perferable, well, it's done. :)

@JeroenDeDauw
Copy link
Member

Is master really preferable? I'd expect the main use case for the sandbox to be that people can just play around with stuff that they can also use on their wiki. For that stable releases are nice. Running master might make the wiki go down some of the time, as not everything is always compatible with each other. (Also depends on which things you run master for... can't do this for all, as composer will say no.jpg. You can do dev versions within the compatible ranges of course.) Having the latest possible dev versions show up is nice for developers and pre release testers. That is quite a different use case though. Given these two downsides for the use case I'm assuming is the main one, I'd not combine these two.

How about

  • sandbox.semantic-mediawiki.org - stable
  • test.semantic-mediawiki.org - dev

@mwjames
Copy link
Contributor Author

mwjames commented Sep 28, 2015

My intake here is that smw.org runs the stable version with examples provided but sandbox.semantic-mediawiki.org runs the dev version which in turn can be used to compare results or allows for quicker intervention.

I do believe that ours tests (at least for SMW core) guard us against any big-boom-cannot-be-installed scenario. Having all extensions in one box makes verification a bit easier when someone comes and says "it doesn't work".

test.semantic-mediawiki.org - dev

I think a bit too much infrastructure, let's keep lean and clean.

@JeroenDeDauw
Copy link
Member

My intake here is that smw.org runs the stable version with examples provided but sandbox.semantic-mediawiki.org runs the dev version which in turn can be used to compare results or allows for quicker intervention.

The main SMW wiki is not intended to be used as a sandbox. Also, the concept of a sandbox is disjoint from running development versions or testing those.

I do believe that ours tests (at least for SMW core) guard us against any big-boom-cannot-be-installed scenario. Having all extensions in one box makes verification a bit easier when someone comes and says "it doesn't work".

For SMW yes. But not for all extensions. If you run SMW master, and there is a breaking change there, it will be incompatible with extensions that relied on the old behaviour, until fixes are made there. This is why the composer dependencies have upper bounds in their version ranges and don't just specify master.

I think a bit too much infrastructure, let's keep lean and clean.

You are right in that this is an additional wiki. And I'm feeling a bit bad for asking for another one to be created. However, I do not agree this is less lean and clean. Single Responsibility Principle. Mashing different concerns into a single class is not more clean, same goes for infrastructure :)

@kghbln
Copy link
Member

kghbln commented Oct 1, 2015

Until 2.3 was released sandbox should stay on master for the simple reason that already Semantic Cite specific content was added. We could then switch to stable as I provide another instance running master. I would call the subdomain "testmaster" since "sandbox" and "test" cannot really be distinguished intuitively by occasional users. Doing it later will also allow me to have more insight into what amount of traffic the current sandbox is getting. This is currently not a very strong VM.

Besides on my private wiki running master I had three fatals within a year, two caused by SMW and one caused by a wmf maintained extension dropping compat for previous versions of MW (well one should not do this anyway if you do not intend to run MW on master also). So there are luckly just a few worries.

@mwjames
Copy link
Contributor Author

mwjames commented Oct 1, 2015

I had three fatals within a year, two caused by SMW and one caused by a wmf maintained extension dropping compat for previous versions of MW

Speaking of the devil ...

image

@kghbln
Copy link
Member

kghbln commented Oct 1, 2015

Hmm, it's working for me. We should hover open an extra issue for reporting troubles with sandbox

@mwjames
Copy link
Contributor Author

mwjames commented Oct 1, 2015

We should hover open an extra issue for reporting troubles with sandbox
of MW (well one should not do this anyway if you do not intend to run MW on master also).

Would be wise I just encountered:

A database query error has occurred. This may indicate a bug in the software.

    Query:
    SELECT 1 FROM `02100_page` WHERE (Array) OR (Array) OR (Array) LIMIT 1
    Function: ApiStashEdit::checkCache
    Error: 1054 Unknown column 'Array' in 'where clause' (localhost)

Caused most likely by https://phabricator.wikimedia.org/T105597

@kghbln
Copy link
Member

kghbln commented Oct 1, 2015

I encountered this issue about two months ago after @thingles updated WikiApiary (WikiApiary/wiki#4) This is quite nasty. I have however not configured sandbox to use redis yet.

@kghbln
Copy link
Member

kghbln commented Oct 1, 2015

The tracking issue for "sandbox.semantic-mediawiki.org" is now at #1186

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@kghbln kghbln

Labels
infrastructure
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
5 participants
@mkroetzsch @JeroenDeDauw @kghbln @mwjames @jongfeli