#1245 Archive wiki.centos.org (read-only)
Closed: Fixed 21 days ago by arrfab. Opened a month ago by arrfab.

Per discussion on centos-devel list in the past, and the PoC to show that it was possible, we'd need to decommission the centos 7 machine used to host https://wiki.centos.org

The goal would be :

  • lock wiki editing
  • convert to html pages
  • upload as first commit to a new git repository
  • modify all content to show that it's a read-only archive
  • had a new VM being use to pull from that git repo and host static files

@dcavalca : as we worked together on the PoC some months ago, what about we kick the official migration ?

Metadata Update from @arrfab:
- Issue tagged with: centos-7-8s-eol, high-gain, medium-trouble

a month ago

Yes, let's get this started. Trying to bring the instance I used for crawling back online so I can do another pass once the wiki is locked.

Ok I have this ready and can kick off another crawl whenever we're ready. Let me know when the wiki is locked for edits and I'll get this started.

well, before we can lock it , we should :

  • prepare the new server that will become https://wiki.centos.org (httpd static pages pulled from git)
  • create that git repo (on git.centos.org or gitlab.com/CentOS ?) and grant permissions for the centos-docs group
  • announce this with a migration date (at least to centos-devel and centos-docs list)

Then we can do it. I can work on the two first items and then we can eventually sync with @shaunm about how/when to announce it ?

@dcavalca , @shaunm : I started (between other tasks) to work on point 1 but for point 2, do you have a preference about where to host the static web pages ? both pagure (git.centos.org) and gitlab will use SSO and group memberships but maybe going with Gitlab makes sense as it's probably easier for quick edit to do it online through web ui, or even MR.
Opinions, thoughts ?

I already have https://gitlab.com/dcavalca/wiki-archive with the previous version (https://dcavalca.gitlab.io/wiki-archive/ is the deployed one, via gitlab pages), if we go with gitlab I'm happy to transfer this to the CentOS org.

Sounds good so let me just create new project and grant you access so that we can push there.
One remark though : it seems that generated static pages are breaking previous links, so indexed links (from search engines) or even our own internal links (website) would not work anymore

Example: https://wiki.centos.org/SpecialInterestGroup when converted doesn't produce https://dcavalca.gitlab.io/wiki-archive/SpecialInterestGroup. Is there a way to remove the generated index with timestamp ?

In parallel just trying https://moinmo.in/HelpOnMoinCommand/ExportDump which just dumps all pages into individual html files and trying if we can just import css/theme and then massive sed -i s/ to add banner

Metadata Update from @arrfab:
- Issue assigned to arrfab

a month ago

I had a little bit of time for this and easy static html files (no theme but content and that's what we asked) tmp link for the example (PoC)
If that works for you we can announce and move on

@dcavalca , @shaunm ^

https://gitlab.com/CentOS/docs/wiki_archives is created and I'll push to that git repository the generated html files once we'll have announced the migration

Tagging @shaunm and @dcavalca again

This looks great to me, thanks! I don't think losing the theme is a big deal. I know some folks had expressed concern about losing the history, but that didn't seem to be salvageable with the crawler approach either.

Went live this morning , from freshly exported content, now pushed to gitlab.com/CentOS/docs/wiki-archives

Metadata Update from @arrfab:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

21 days ago

Login to comment on this ticket.