Aufenthaltsraum Server Maintenance Feed

Upgrade's over.

Time’s around 21:15 and the upgrade is successfully completed – we’re ready to roll. Please check your services.

Say hi to Xenial Xerus, which even appears to be an LTS release – so we may theoretically live without reboots for around 5 years. (Well, a kernel update now and then can’t harm, though.)

(By the way, the upgrade needed moar disk space, so $ADMIN has “vacuumed” some journald space. We now have around 1.7GB of disk space after the upgrade.)

First reboot cycle is over!

It’s not even 20:30 UTC, but today’s first reboot cycle is over. Not much has happenned, though: The system was shut down to take a snapshot.

The actual distro update is on its way now, which will cause a another reboot later.

(By the way, on this occasion $ADMIN noticed that there is less than 3GB left on our disk, so please mind that.)

Brace for impact!

It’s now around 19:50 UTC. The upgrade process will begin in less than 15 minutes. Cross your fingers and brace for impact!

Upcoming Distro Upgrade

The Linux distro on the server will be upgraded on 2016-05-15. The upgrade process will start at around 20h UTC.

Whole data on the server will be backed up on this occasion and there will be at least two reboots. If everything goes well, it will be all over until midnight (UTC) and the services will be capable of running as usual.

On the day of the upgrade there will be “live” updates in this feed (as long as the server is up).

Our CI vs. GitLab CI

$ADMIN is considering deprecating our self-hosted CI server in favor of GitLab CI because the CI server uses 80.8MB (~16%) of our (precious) RAM (in idle mode!). (And GitLab CI is probably technically superior than the service we are hosting.)

To make a possible migration easier, here is already a tutorial-like wiki page describing how to use GitLab CI for LaTeX projects. is inaccessible. is now showing some domain parking nonsense (thanks to ML for noticing this), and $ADMIN could not found any notice from the registrar which explains why this is the case (except for the fact that the domain has also disappeared from the domain management interface). $ADMIN will be looking into this – #technikeristinformiert.

You can use (or while this problem persists, but beware that these domains are also managed by the same registerar. Alternatively, the IP addresses and 2604:a880:400:d0::b:e001 of our server are likely to be unchanged for a while.

The wiki is online again.

The self-hosted wiki mirror is online again. It is, however, not a full-blown Gollum instance anymore, but just a bunch of poorly styled HTML pages. It does what it should be doing, though.

Bonus: Here is the crappy Ruby script which generates the pages:

require 'rubygems'
require 'gollum-lib'
require 'fileutils'

wiki ="../")

preprefix = "<html><head><meta http-equiv=\"Content-Type\"" +
  " content=\"text/html; charset=utf-8\"><title>Aufenthaltsraum Wiki: "
prefix = "</title></head><body>"
suffix = "<hr><a href=\"/Home\">Home</a> | " +
  "<a href=\"/pages.html\">List of Pages</a></body></html>"

page_list = preprefix + "List of Pages" + prefix + "<ul>"

wiki.pages.each do |p|
  path = p.escaped_url_path
  title = p.title
  page_list += "<li><a href=\"" + path + "\">" + title + "</a></li>"
  FileUtils::mkdir_p path
  File.write(path + '/index.html',
             preprefix + title + prefix + p.formatted_data + suffix)

page_list += "</ul>" + prefix
File.write("pages.html", page_list)

The wiki is down.

The self-hosted mirror of our wiki will be down for a while while $ADMIN is trying out some less momory-hungry alternatives. Expect web server restarts.

Hello, world!

The maintenance feed is now online and will inform you about non-critical maintenance actitivities on the server which don’t require a reboot and aren’t expected to effect user space activity in any other way.

There is also an atom feed available.