These installation notes are mostly a note to myself, documenting the installation process of a Gitlab Omnibus Container in Docker, plus Gitlab Runners.

OS Setup

We are installing into /export/gitlab, a 10G xfs slice from the local flash pool:

# lvcreate -n gitlab -L 10G data
# mkfs -t xfs /dev/data/gitlab
# mkdir /export/gitlab
# mount /dev/data/gitlab /export/gitlab
# echo "/dev/data/gitlab\t/export/gitlab\txfs\tbsdgroups,usrquota,grpquota,attr2,nofail,noatime 1 2" >> /etc/fstab

# mkdir /export/gitlab/{gitlab,gitlab-runner}
# mkdir /export/gitlab/gitlab/{config,data,logs}

Docker

We are using docker-compose to run this, with a .env (dotenv) like so:

At work, I am in an ongoing discussion with a number of people on the Observability of Outliers. It started with the age-old question “How do I find slow queries in my application?” aka “What would I want from tooling to get that data and where should that tooling sit?”

As a developer, I just want to automatically identify and isolate slow queries!

Where I work, we do have SolarWinds Database Performance Monitor aka Vividcortex to find slow queries, so that helps. But that collects data at the database, which means you get to see slow queries, but maybe not application context.

I have been asked to document my home sensor network. Being married to a person with a background in web security sets boundary conditions:

1. No cloud. We are running all services locally.
2. No control, only metrics.

I am collecting data from a number of plugs with power meters over Wi-Fi, using the MQTT protocol. I am also collecting data from a number of temperature sensors over Zigbee, and convert to MQTT. The MQTT data is ingested into Influx, and then read and plotted in Grafana. All of this is dockered and runs locally on an Ubuntu server.

Es gibt ein Interview mit Stefan Ramesohl vom Umweltministerium (des Bundes) in Netzpolitik.org: “Warum niemand weiß, wie viele Rechenzentren es in Europa gibt ”. Im Wesentlichen hat das Umweltministerium angesagt, daß es auf europäischer Ebene Rechenzentren erfassen und katalogisieren will, um in einem zweiten Schritt den Energieverbrauch von Rechenzentren zu regulieren.

Das ist sehr spannend, denn derzeit gibt es keine Übersicht über Rechenzentren in Europa, und tatsächlich sind einige Rechenzentrumsbetreiber sehr paranoid, was den genauen Standort ihrer Hardware angeht und wieviel und welche Hardware darin ist oder was diese tut. Das ist zwar lächerlich - es ist sehr schwierig eine Energiesenke wie ein Rechenzentrum und ihre Abwärme zu verstecken - aber auch ein sehr sensitives Thema.

Sometimes things change in a way that is hard to put a finger on, but I am doing this MySQL thing since 3.23, and commercially since 2005, and the environment is changing. These days, when you talk to people in need of MySQL, the first thing you have to ask them is “Which MySQL”. And by that I do not mean a version number in the first place.

The answer may be:

Sven Geggus trolled me . A bunch of nerds were speaking about what’s wrong with biking in Germany, and he wrote:

Tweet : Der @isotopp wohnt doch in Holland. Wo sind denn bei euch die Vorfahrtsregeln anders und könnte man da was sinnvoll für .de übernehmen? – @isotopp is living in the Netherlands. So how is the right of way different and how could .de learn from this?

Biking in the Netherlands does not suck, and that is not because of any specific traffic rules being any different, but because traffic is fundamentally different, and that somehow escalated.

Ok, so the title is a bit of an exaggeration, I am about 80% done with the main games quests. Still, the shape and feel of the story is firmly established and it is fascinating. I have some 110 hours in “Assassins Creed: Origin”, and some 65 hours or so in “Assassins Creed: Odyssey”.

You can play “Assassins Creed: Odyssey” as Alexios or Kassandra, and the change is permanent for the playthrough for storytelling reasons. My recommendation is that you play as Kassandra, see below.

Back when I was still commuting to work in an office, in the far past, I used to be in the Spaces building in Vijzelgracht in Amsterdam, every day. Mark Wagenbuur of BicycleDutch has been there today, and tweeted :

Amsterdam is changing dramatically. I was in Vijzelstraat for the first time in about 2 years and I was pleasantly surprised! Looked up the old situation in Google StreetView for some before and afters.

At work, replication chains have a single primary database node, to which you write, and then multiple replicas, in multiple AZs.

Here is what the one sample chain looks like in Orchestrator:

instance-918d is the current primary, in the blue AZ. Replicas in orange and green are in other AZs. Blue badges indicate multiple replicas, eg (38) means 38 machines.

When you talk to a database, you get two database handles:

Ich schrieb in einem Twitter Thread über Posix Dateisysteme vs. Object Stores :

UNIX FS ist 1974. BSD FFS ist 1984. XFS ist 1994. ZFS (und Btrfs und Wafl) sind LFS, also 2004. Object Storages, LSM, “RocksDB” ist ca. 2014, um den Takt zu halten.

und wurde gefragt: “Was kommt 2024”. Meine halb spöttische, halb ernst gemeinte Antwort war:

Irrelevant.

2024 läuft Dein Code serverless bei einem professionellen Betreiber und vom lokalen System und dem lokalen Dateisystem kriegst Du nix mehr zu sehen außer einer monatlichen Rechnung.