[AFTER 2026-08-04] Remove the remotebuild user, and ssh access for the root user

[jfly]

If you ended up here by clicking link on makemake, please stop using the remotebuild and root users in favor of your own personal unix account. They were created in #33. If your user is missing or misconfigured, please file an issue or send in a PR.

---

In #33 we added unix accounts for every human (and one bot) with access to makemake, and deprecated the remotebuild and root users.

However, those remotebuild and root users still exist and have ssh keys. After an appropriate transition period, let's:

• Remove the remotebuild user (and clean up /home/remotebuild).
• Remove the ssh authorized keys for the root user.

[jfly]

Here's a helper script to visualize logins over time:

plot.py:

#!/usr/bin/env nix-shell
#! nix-shell -i python
#! nix-shell -p 'python3.withPackages (ps: [ps.pandas ps.matplotlib])'

# To use this, feed the logs via stdin. Something like this:
# ssh makemake 'sudo journalctl SYSLOG_IDENTIFIER=systemd --grep="Started Session .* of User" --output=short-iso' | ./plot.py

from typing import cast
import fileinput
import pandas as pd
import matplotlib.pyplot as plt
import re
import datetime as dt

# Login events (`journalctl --output=short-iso`) look like this:
# 2026-07-29T02:33:22+02:00 makemake systemd[1]: Started Session 1092 of User jfly.
LOGIN_RE = re.compile(r"^(?P<ts>\S+).* of User (?P<user>\S+)\.$")

def parse_logs() -> pd.DataFrame:
    rows = []
    for line in fileinput.input(encoding="utf-8"):
        if line.startswith("-- Boot "):
            continue

        match = LOGIN_RE.match(line)
        assert match is not None, f"Invalid line: {line}"
        groups = match.groupdict()

        time = dt.datetime.fromisoformat(groups["ts"])
        rows.append({
            "time": time,
            "date": time.date(),
            "user": groups["user"],
        })

    return pd.DataFrame(rows)

def plot():
    logins_df = parse_logs()

    # Count of logins per (date, user).
    counts = logins_df.groupby(["date", "user"]).size().unstack(fill_value=0).sort_index()
    ax = counts.plot(kind="bar", stacked=True, colormap="tab20")

    x_pos: float = cast(float, counts.index.get_loc(dt.date(2026, 7, 28))) + 0.5
    ax.axvline(x_pos, color="red")
    ax.text(
        x_pos,
        ax.get_ylim()[1]*0.95,
        "Introduction of multiple accounts",
        rotation=90,
        va="top",
        ha="right",
        color="red",
    )

    plt.show(block=True)

plot()

To run it, feed logs via stdin. Something like this:

$ ssh makemake 'sudo journalctl SYSLOG_IDENTIFIER=systemd --grep="Started Session .* of User" --output=short-iso' | ./plot.py

You'll see something like this: