Skip to content

fix(templates): the terminal was not the whole of that refusal - #124

Merged
aaldersondev merged 1 commit into
mainfrom
fix/steam-cold-call
Aug 25, 2026
Merged

fix(templates): the terminal was not the whole of that refusal#124
aaldersondev merged 1 commit into
mainfrom
fix/steam-cold-call

Conversation

@aaldersondev

Copy link
Copy Markdown
Contributor

What happened

v0.19.2 shipped a diagnosis saying Missing configuration means the install container had a terminal, and telling the operator to update their node.

An installation then ran on a node already carrying that fix — daemon 0.19.2, commit a9faf43, TERM unset inside the container — and its first attempt refused with exactly that message. The second downloaded all 6.87 GB and the server installed.

So the message sent somebody to check a version that was already right. That is the worst kind of diagnosis: precise, confident, and pointing away from the fix.

What was actually measured, and what was assumed

Measured, and it stands: a terminal fails every time — seven out of seven with one, three out of three without, interleaved so neither time nor a Steam outage explains it. A warm-up +quit before the real call does not help, and no redirection inside the script escapes it. Tty: false stays.

Assumed, and wrong: that a cause found first was the only one. With no terminal at all, a cold SteamCMD still refuses its first call now and then — twice in the runs watched, once on the node above.

What changes

Only what the script says when it gives up:

SteamCMD logged in and then refused the app configuration,
three times, without downloading anything. Two things do that:

  - A terminal on the install container. The panel stopped
    giving it one in Hopper 0.19.2, and against an older node
    no number of attempts gets past it. Check that first.

  - A cold SteamCMD refusing its first call. That one the next
    attempt usually gets through, which is what these three
    were for. Reinstall in a few minutes.

The comment above the retry loop and docs/templates.md carry the same correction, and the daemon's own docblock no longer claims that line decides whether Steam installs at all — it takes down a wall, and the retry handles what is left standing.

The retry that carried the installation through was already there. This is what it was for.

🤖 Generated with Claude Code

v0.19.2 said `Missing configuration` meant the install container had a
terminal, and told operators to update their node. An installation on a node
already carrying that fix — daemon 0.19.2, `TERM` unset in the container,
verified — refused its first call with exactly that message and downloaded
all 6.87 GB on the second.

So the message sent people to check a version that was already right, and
that is the worst kind of diagnosis: precise, confident, and pointing away
from the fix. It now names both causes and says which one a retry can get
past.

The tty finding stands as measured — seven failures out of seven with one,
three successes out of three without, and a warm-up `+quit` beforehand does
not help — and `Tty: false` stays. What was wrong was treating a cause found
first as the only one.

Nothing else changes: the retry that carried this installation through was
already there, and this is what it was for.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ReX2JSG6VKYShanE1F69KP
@aaldersondev
aaldersondev merged commit 87af0d0 into main Aug 25, 2026
3 checks passed
@aaldersondev
aaldersondev deleted the fix/steam-cold-call branch August 25, 2026 14:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant