Track Firebase hosting migration

[petebacondarwin]

• @Narretz setup errors.anguarjs.org firebase redirects
• @IgorMinar double check all certs
• @petebacondarwin double check the old ngix config to ensure we haven't missed anything
• @Narretz resolve code.angularjs.org timeout issues for large files
• @IgorMinar update dns to point to the new sites
• @IgorMinar update blog.angularjs.org to point to blogger or blog.angular.io
• @petebacondarwin shut down old ci.angularjs.org jobs that were keeping the site up to date
• @Narretz remove old deployment code from angularjs.org project
• @petebacondarwin update hosting details doc
• migrate ci.angularjs.org
  • @petebacondarwin obtain and install a cert on the machine
  • @IgorMinar update dns details ci.angularjs.org
• @petebacondarwin shut down old cloud VMs
• @Narretz follow up on firebase quote limit issues in order to ensure that we don't get shut down
• @Narretz code.angularjs.org robots.txt issues (see Track Firebase hosting migration #16417 (comment))
• @Narretz docs.angularjs.org sitemap and robots.txt issues (see google.com search results are linking doc pages without stylesheets #16432 (comment))
• @Narretz restore escaped fragment indexing google.com search results are linking doc pages without stylesheets #16432

[IgorMinar]

@petebacondarwin thank you!

[IgorMinar]

@Narretz is it correct that the timeout issue is resolved?

[Narretz]

@IgorMinar yes the issue is resolved. It was an encoding issue not a file size issue. The zip files were additional gzipped for faster upload but this confused gcs / Firebase. I replaced all the available zips with their default form and now it works. The deployment has been updated too

[IgorMinar]

So I guess that means that we are good to go then. Right?

I book Wed 9-10am PT on my calendar to make this go live + sent out an invite to you two.

[IgorMinar]

@petebacondarwin what about blog.angularjs.org? have you thought about it? if we don't change anything, I believe that it will break because of the certs and the current proxy setup.

I think we'll need to do something like:

• https://support.google.com/domains/answer/6069231?hl=en
• and https://www.howtoshout.com/enable-https-blogger-blog-custom-domain/

[petebacondarwin]

@IgorMinar - yes we need to setup the DNS to CNAME to point to blogger and then set up the blog on Blogger to accept incoming connections from that domain - I believe there is now a setting that allows HTTPS. I don't have admin rights to do either of these two.

The alternative is to change blog.angularjs.org to point directly to blog.angular.io, which is not hosted on blogger but something else...

[IgorMinar]

I was thinking about this more and I think the lowest risk will be to just setup a redirect from blog.angularjs.org to blogger. The current DNS setup is messed up anyway and most urls on the web point to either https://blog.angularjs.org or http://angularjs.blogspot.com/<somepath>.html.

[petebacondarwin]

@IgorMinar that would be fine by me but I think it will need, yet another, Firebase project setup for the subdomain, right?

[IgorMinar]

all but ci.angularjs.org are switched now. @petebacondarwin we need to decide what to do with the jenkins instance. without the global proxy we no longer have a signed certificate for it. Could you look into getting a cert from lets encrypt and installing it on the box?

[IgorMinar]

haha.. so much for "we need to decide". above is my proposal. the only two alternatives is to use a self-signed cert (bad ux) or switch to http (bad security). @petebacondarwin if there are other options please propose them.

[petebacondarwin]

Can we continue to use the IP address directly for Jenkins, @IgorMinar?