Nov 6, 2022


This week I read:

Roleplaying Games


Another session, another postponement.

The previous session was on the 5th of October, we’re supposed to play every first Wednesday and third Tuesday of the month, and so the next session is, in principle, on the 15th. A few of the players even missed the previous session, and so haven’t played since the 16th of August.

There are a couple of players who I suspect don’t actually want to play, but just keep declining sessions rather than say that. One player said in advance that the schedule doesn’t really work for him, and so he’d only be able to make the Wednesday sessions, which wasn’t a great start. But about half of the players did all seem pretty keen and able to make every session: they’ve just often had something better to do.

Obviously, D&D isn’t the most important thing in the world. There are legitimate reasons to miss sessions, I’ve missed sessions, it’s not really a problem unless it becomes a habit, and with this group, it has.

So next week I need to have a chat with the group, to see who’s actually interested in playing (if anyone), and to make any reasonable changes I can to help with attendance. For example, maybe attendance will just be so sporadic that any sort of story is infeasible, in which case maybe we should play a megadungeon and make sure every session ends with the PCs returning to their home base. Or maybe we change to an open table, and I have a stack of pregens available for interested newbies, and we play with whoever shows up on the night. Or maybe the players take on scheduling, and just let me know a few days in advance of when they’d like to play.

It would be a shame to lose the group entirely, but right now we’re just not playing.

Cartographic Curiosities

This game, on the other hand, is doing great. We’ve settled into a rhythm of playing every other week: in principle we’re rotating amongst 4 games, but two of the GMs don’t often schedule something.

I’ve adopted a faction system, based on this blog post, though I’ve tweaked it a bit (if it ends up working out well, I’ll write something up). A few things have happened, but the characters have been out wandering the wilds and so haven’t seen the results yet: but when they return to the main town where they have all their money, they’ll hear talk of a new military order established by the local leader to vanquish “the forces of chaos” which is preparing for its first expedition, and they’ll also be approached by a bumbling spy (a newly-established agent of one of those very forces) who is not-very-subtly going to try to recruit them to spy on a mutual enemy. I hope the players adopt a “the enemy of my enemy is my friend” attitude with this guy.

The players got some background lore on one of the big tensions in the setting: the frost elves and their malevolent leader, the Cold Prince, who were magically banished some ~900 years ago. They’ve got snippets of the story from various sources and so, this session, when they encountered a wise old scholar, they got him to fill in the missing details. This was actually brilliant, and totally unplanned, as the night before the session I was wondering how to get across those details, since they’re not just history: they’re adventure hooks. And then the players went and just asked!

But it’s not all big forces and world-ending threats, we also spent about half an hour in a conflict between two factions of michievous sprites, with each side offering the PCs more and more elaborate bribes to go and mercilessly slaughter the other. A little bit of whimsy to break up the serious parts goes a long way.

Wicked Ones

  1. Grow the hoard
  2. Start the Hobbit war
  3. Ally with the orc camp against the martial-arts monastery
  4. ???
  5. ???
  6. Profit!

That’s the nefarious master plan, 7 sessions into the campaign. Last session the players completed step 2, this session an important first step towards step 3 was made: having talks with the orc leader and agreeing to a combined raid on the monastery, which is by itself stronger than either the dungeon or the orc camp individually. We also had a major dungeon invasion this session, in which three adventurers overcame the defenses with relative ease and made their way into the sanctum for a pitched battle against the PCs.

The book says each step of the plan should take 2 or 3 sessions to complete, so we’re actually right on track so far. If we’re now 1 session into step 3, the campaign could be over in as few as 7 or as many as 11 more sessions.

I quite like the game, some parts still trip me up—remembering to do XP at the end of each session is surprisingly hard, and even when intentionally trying to inflict harsher consequences I don’t seem to do a good job of actually putting the PCs in serious risk—but it’s fun. I think it’s about time to do a little mid-campaign check-in and get some focussed feedback from the players: they also seem to like it, but people are notoriously bad at expressing opinions.

I am glad that it’s designed for short-to-medium length campaigns though, because it’s not my preferred style of game: fun to play for a few months, then back to something more traditional like Traveller or Call of Cthulhu for a while.


OCI containers on NixOS

NixOS has some handy options for running OCI containers as systemd units, but docker has a couple of significant footguns:

  • It’s too easy to delete named volumes, even ones which are actively being used by a running container.

  • Forwarded ports are bound to all interfaces by default, bypassing iptables and allowing anything on the network to communicate with your container.

I solved those issues through convention: all volume mounts are bind-mounts to the host filesystem and not named volumes, and all ports are explicitly bound to

But I also often have two containers which I want to be on the same local network (eg, a service and its database), and the NixOS module doesn’t have a convenient way to do that, I have to add a pre-start step to the container’s systemd unit to create the network (if it doesn’t already exist) and pass --network=FOO to the docker command:

virtualisation.oci-containers.containers.bookdb = {
  /* ... */
  extraOptions = [ "--network=bookdb_network" ];
  dependsOn = [ "bookdb-db" ];

virtualisation.oci-containers.containers.bookdb-db = {
  /* ... */
  extraOptions = [ "--network=bookdb_network" ];

systemd.services.docker-bookdb-db.preStart = "docker network create -d bridge bookdb_network | true";

If I want to pull the container on start (handy for :latest images), that’s another pre-start step too:

systemd.services.docker-bookdb.preStart = "docker pull ${image}";

So, I had some configuration practices which I manually applied everywhere I used containers, rather than just writing a better abstraction which did all that for me. So this weekend I made that better abstraction.

Not a very complex abstraction, but it addresses my needs:

  • All volumes are either bind-mounts to the host filesystem, or named volumes which get automatically bind-mounted to a predictable location on the filesystem.
  • All ports are implicitly bound to
  • If a network name is given, it’s created when the container starts, and the container is added to the network, automatically.
  • Images get pulled on start (by default).
  • Containers start at boot (by default).

I also added a check to my lint script to forbid use of the standard NixOS OCI containers options outside of the module implementing this new abstraction, to make sure everything is done right.

  1. Indeed, if the collection was arranged in publication order, The Silver Spike would be second.↩︎