Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
Dear all, dear Tim, I want to extend the thank you from the Uzbek Wikipedia community. I let them use their own words (although, translated for the benefit of this list). user 1: wow, it's great! who has done it user 2: wikimedia programmers user 1; super! real heroes! user 3: thank you, thank you, thank you! user 4: so now we're waiting for google to index uzwiki user 2: it's awesome, i cannot even express how awesome it is user 2: Thank you guys for helping us. And special thanks to Tim Starling. user 2: We're going to introduce Tim Starling day in Uzbek Wikipedia as well I also pinged my contacts at Google, hoping that they can schedule the reindexing soon. Thanks to everyone involved. Rahmat va salomlar! On Mon, Jan 7, 2013 at 6:17 AM, Tim Starling tstarl...@wikimedia.orgwrote: On 24/12/12 20:23, Anonymous User wrote: I don't know how much effort each of these two measures would be. If you'd ask me, I would suggest to be very serious, but we are not under a deadline (the situation has been like this for more than a year now), and setting the rel=caonical would already be really, really helpful. This is done now. It would be good if Google could crawl uz.wikipedia.org to update the canonical URLs. In case anyone is wondering, I don't think this would be a good thing to do on zh.wikipedia.org. The Chinese government would happily block *.wikipedia.org port 443 if it became popular. At least the current situation provides a way to work around keyword filtering for people who are sufficiently motivated -- if HTTPS was blocked, it would be much less useful. -- Tim Starling ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On 24/12/12 20:23, Anonymous User wrote: I don't know how much effort each of these two measures would be. If you'd ask me, I would suggest to be very serious, but we are not under a deadline (the situation has been like this for more than a year now), and setting the rel=caonical would already be really, really helpful. This is done now. It would be good if Google could crawl uz.wikipedia.org to update the canonical URLs. In case anyone is wondering, I don't think this would be a good thing to do on zh.wikipedia.org. The Chinese government would happily block *.wikipedia.org port 443 if it became popular. At least the current situation provides a way to work around keyword filtering for people who are sufficiently motivated -- if HTTPS was blocked, it would be much less useful. -- Tim Starling ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
Thank you everyone for your help, and I especially appreciate Sumanah for setting up the bug and making the RT ticket happen. Unfortunately, I cannot track the progress of the RT ticket, but I am sure someone will keep us posted on the Bugzilla bug or on this list. Eternal gratitude to everyone involved! Regarding the discussion about the IP, here are my thoughts, but maybe I am missing a number of the technical intricacies here: I am not sure if I understand correctly, but having access through the IP does not seem to improve the situation. Remember that access through https is perfectly possible. So for anyone who really wants to access the wiki and knows about the https-access, they can. The problem are those people that do not know about it. There are no people coming to the Uzbek Wikipedia through search engines, because they link to the http-protocol version of the URLs. That is why making the https-Version canonical and thus the addresses the search engines eventually point to much more useful. One could make the IP-Adress directly canonical -- otherwise setting up this extra-IP would not bring any advantage -- and thus avoid DNS, but this is not required to resolve the current problem, and it sure would not be very pretty. Also, IP-banning a single IP through an ISP is all but hard for an ISP. And anyway, it would need to be configured to setup a canonical URL, which is the currently suggested next step anyway. Therefore I would suggest to refrain from setting up a specific IP (besides the technicality of having too few IPs around), at least for now, and concentrate on canonizing the HTTPS protocol (as summed up in the Bug). Thank you again! On Fri, Dec 28, 2012 at 3:05 AM, John Vandenberg jay...@gmail.com wrote: How many languages _need_ this? Is it only one language-project? If you only need one IP address, to avoid censorship by one country, it should be achievable. John Vandenberg. sent from Galaxy Note On Dec 28, 2012 4:21 AM, Leslie Carr lc...@wikimedia.org wrote: I wish that http://208.80.154.225/wiki/Bosh_Sahifa and https://208.80.154.225/wiki/Bosh_Sahifa would work, too, but the foundation apparently can't or chooses not to afford separate IP addresses for each language's Wikipedia. As one of the network folks, I will answer this. We do not have enough public IP(v4)s for an address for each language in each project, and unless someone gives us a major donation of IPv4 addresses (anyone have a spare /20 laying around?), I don't think we will be able to make this happen as we are frugal with our existing IPs and the allocating authorities (RIPE and ARIN) are being quite strict with their new IPv4 allocations. If you'd like to read more about IP allocation policies, here's a few links https://www.arin.net/policy/nrpm.html#four3 https://www.arin.net/resources/request/ipv4_depletion.html https://www.ripe.net/ripe/docs/ripe-553 (see section 5.6) Leslie -- Leslie Carr Wikimedia Foundation AS 14907, 43821 http://as14907.peeringdb.com/ ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On 27/12/2012 5:54 AM, Anonymous User wrote: thank you again for your answers so far. I would have had hoped to have more voices participating, but everyone who did agreed that it should be done. I think this is the closest I've ever seen to universal support on Wikimedia-l ever. :-) -- Coren / Marc ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On 12/27/2012 10:46 AM, Marc A. Pelletier wrote: On 27/12/2012 5:54 AM, Anonymous User wrote: thank you again for your answers so far. I would have had hoped to have more voices participating, but everyone who did agreed that it should be done. I think this is the closest I've ever seen to universal support on Wikimedia-l ever. :-) -- Coren / Marc So, I just asked Chris Steipp (WMF engineer in charge of software security) for his thoughts on this: I can add a, I think it's a good idea to the list, but Ops will need to be ok with the shift. I don't think it would be a problem, but it does mean google spidering our https site, and that may concern them. I think ops would also be the ones to implement the actual change. So in my opinion we can move discussion over to https://bugzilla.wikimedia.org/show_bug.cgi?id=43466 (when serving Uzbek Wikipedia, make HTTPS canonical). I've asked a bug wrangler to contact Ops about it as well. -- Sumana Harihareswara Engineering Community Manager Wikimedia Foundation ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On 12/27/2012 03:08 PM, Sumana Harihareswara wrote: So in my opinion we can move discussion over to https://bugzilla.wikimedia.org/show_bug.cgi?id=43466 (when serving Uzbek Wikipedia, make HTTPS canonical). I've asked a bug wrangler to contact Ops about it as well. (Maybe I should just include a link to https://meta.wikimedia.org/wiki/Glossary in every email I send because I do use a lot of jargon! Sorry about that.) -- Sumana Harihareswara Engineering Community Manager Wikimedia Foundation ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
I wish that http://208.80.154.225/wiki/Bosh_Sahifa and https://208.80.154.225/wiki/Bosh_Sahifa would work, too, but the foundation apparently can't or chooses not to afford separate IP addresses for each language's Wikipedia. As one of the network folks, I will answer this. We do not have enough public IP(v4)s for an address for each language in each project, and unless someone gives us a major donation of IPv4 addresses (anyone have a spare /20 laying around?), I don't think we will be able to make this happen as we are frugal with our existing IPs and the allocating authorities (RIPE and ARIN) are being quite strict with their new IPv4 allocations. If you'd like to read more about IP allocation policies, here's a few links https://www.arin.net/policy/nrpm.html#four3 https://www.arin.net/resources/request/ipv4_depletion.html https://www.ripe.net/ripe/docs/ripe-553 (see section 5.6) Leslie -- Leslie Carr Wikimedia Foundation AS 14907, 43821 http://as14907.peeringdb.com/ ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
Leslie Carr lc...@wikimedia.org schrieb: I wish that http://208.80.154.225/wiki/Bosh_Sahifa and https://208.80.154.225/wiki/Bosh_Sahifa would work, too, but the foundation apparently can't or chooses not to afford separate IP addresses for each language's Wikipedia. As one of the network folks, I will answer this. We do not have enough public IP(v4)s for an address for each language in each project, and unless someone gives us a major donation of IPv4 addresses (anyone have a spare /20 laying around?), I don't think we will be able to make this happen as we are frugal with our existing IPs and the allocating authorities (RIPE and ARIN) are being quite strict with their new IPv4 allocations. If you'd like to read more about IP allocation policies, here's a few links https://www.arin.net/policy/nrpm.html#four3 https://www.arin.net/resources/request/ipv4_depletion.html https://www.ripe.net/ripe/docs/ripe-553 (see section 5.6) Just an idea, which is not very beautiful: What about a router forwarding ports to the correct machine by using iptables? Would that also work in connection with search engines? Cheers Marco ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On Thu, Dec 27, 2012 at 1:39 PM, Marco Fleckinger marco.fleckin...@wikipedia.at wrote: Leslie Carr lc...@wikimedia.org schrieb: I wish that http://208.80.154.225/wiki/Bosh_Sahifa and https://208.80.154.225/wiki/Bosh_Sahifa would work, too, but the foundation apparently can't or chooses not to afford separate IP addresses for each language's Wikipedia. As one of the network folks, I will answer this. We do not have enough public IP(v4)s for an address for each language in each project, and unless someone gives us a major donation of IPv4 addresses (anyone have a spare /20 laying around?), I don't think we will be able to make this happen as we are frugal with our existing IPs and the allocating authorities (RIPE and ARIN) are being quite strict with their new IPv4 allocations. If you'd like to read more about IP allocation policies, here's a few links https://www.arin.net/policy/nrpm.html#four3 https://www.arin.net/resources/request/ipv4_depletion.html https://www.ripe.net/ripe/docs/ripe-553 (see section 5.6) Just an idea, which is not very beautiful: What about a router forwarding ports to the correct machine by using iptables? Would that also work in connection with search engines? Are you suggesting we use different nonstandard ports for each different wiki/language combo that resides on the same IP ? Cheers Marco ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l -- Leslie Carr Wikimedia Foundation AS 14907, 43821 http://as14907.peeringdb.com/ ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
Leslie Carr lc...@wikimedia.org schrieb: On Thu, Dec 27, 2012 at 1:39 PM, Marco Fleckinger marco.fleckin...@wikipedia.at wrote: Just an idea, which is not very beautiful: What about a router forwarding ports to the correct machine by using iptables? Would that also work in connection with search engines? Are you suggesting we use different nonstandard ports for each different wiki/language combo that resides on the same IP ? Yes exactly! ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On Thu, Dec 27, 2012 at 2:37 PM, Marco Fleckinger marco.fleckin...@wikipedia.at wrote: Leslie Carr lc...@wikimedia.org schrieb: On Thu, Dec 27, 2012 at 1:39 PM, Marco Fleckinger marco.fleckin...@wikipedia.at wrote: Just an idea, which is not very beautiful: What about a router forwarding ports to the correct machine by using iptables? Would that also work in connection with search engines? Are you suggesting we use different nonstandard ports for each different wiki/language combo that resides on the same IP ? Yes exactly! I guess that is theoretically possible with a more intrusive load balancer in the middle. We need the HOST information from the http header to be added as we have our varnish caches serving multiple services, not one(or more) per language/project combo. I'm pretty sure that lvs doesn't have this ability (which we use). Some large commercial load balancers have the ability to rewrite some headers, but that would be a pretty intensive operation (think lots of cpu needed, since it needs to terminate SSL and then rewrite headers) and would probably be expensive. If you have another way you think we can do this, I am all ears! We may want to move this discussion to wikitech-l as all the technical discussions probably bore most of the people on wikimedia-l Leslie ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l -- Leslie Carr Wikimedia Foundation AS 14907, 43821 http://as14907.peeringdb.com/ ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
How many languages _need_ this? Is it only one language-project? If you only need one IP address, to avoid censorship by one country, it should be achievable. John Vandenberg. sent from Galaxy Note On Dec 28, 2012 4:21 AM, Leslie Carr lc...@wikimedia.org wrote: I wish that http://208.80.154.225/wiki/Bosh_Sahifa and https://208.80.154.225/wiki/Bosh_Sahifa would work, too, but the foundation apparently can't or chooses not to afford separate IP addresses for each language's Wikipedia. As one of the network folks, I will answer this. We do not have enough public IP(v4)s for an address for each language in each project, and unless someone gives us a major donation of IPv4 addresses (anyone have a spare /20 laying around?), I don't think we will be able to make this happen as we are frugal with our existing IPs and the allocating authorities (RIPE and ARIN) are being quite strict with their new IPv4 allocations. If you'd like to read more about IP allocation policies, here's a few links https://www.arin.net/policy/nrpm.html#four3 https://www.arin.net/resources/request/ipv4_depletion.html https://www.ripe.net/ripe/docs/ripe-553 (see section 5.6) Leslie -- Leslie Carr Wikimedia Foundation AS 14907, 43821 http://as14907.peeringdb.com/ ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On Mon, Dec 24, 2012 at 1:15 AM, Tim Starling tstarl...@wikimedia.org wrote: Is it enough to set the link rel=canonical, or is it also necessary to redirect? When I asked the nice folks at Google's search team, they answered me the following: * The best answer would for them to use rel=canonical tags so that http://example.wikimedia.uz points to http*s*://example.wikimedia.uz. So I'd send them this page: http://support.google.com/webmasters/bin/answer.py?hl=enanswer=139394 and tell them to start doing that. If they're very serious (and it's a small property, so there's not much risk) then they could make every http page 301 to the https version as well. I don't know how much effort each of these two measures would be. If you'd ask me, I would suggest to be very serious, but we are not under a deadline (the situation has been like this for more than a year now), and setting the rel=caonical would already be really, really helpful. Thank you all for your encouraging comments so far. ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l
Re: [Wikimedia-l] No access to the Uzbek Wikipedia in Uzbekistan
On 23/12/12 22:15, Anonymous User wrote: We discussed both with search engine providers and Wikimedia developers if there is a way to resolve this issue, and there is: by making the HTTPS version of the Uzbek Wikipedia canonical the search engines would list the HTTPS version in the search results, thus circumventing the glitch. As far as I understood the technical folks at Wikimedia this can be done with a small amount of effort. Is it enough to set the link rel=canonical, or is it also necessary to redirect? Either way, the Squid cache would have to be purged, then the search engines would have to reread that site. -- Tim Starling ___ Wikimedia-l mailing list Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l