8000 feat(osx): run PostgreSQL 16 for the Mac's fleet account, yielding to its owner by inkatze · Pull Request #250 · inkatze/dotfiles · GitHub
[go: up one dir, main page]

Skip to content

feat(osx): run PostgreSQL 16 for the Mac's fleet account, yielding to its owner - #250

Merged
inkatze merged 23 commits into
mainfrom
planwright/flight/mc-dotfiles-ad-hoc-3117da64
Nov 14, 2026
Merged

inkatze merged 23 commits into
mainfrom
planwright/flight/mc-dotfiles-ad-hoc-3117da64

Conversation

@inkatze
@inkatze inkatze commented Nov 12, 2026 •
Copy link
Copy Markdown
Owner

Guard change for review

roles/osx/files/runner/postgres-gate.sh is exempted from the "no PostgreSQL configuration" check in scripts/services-declaration-test.sh, which now scans it with a stricter, file-specific pattern. The gate passes listen_addresses=localhost as a command-line option and writes no config file; the exemption allows only that exact value and still rejects pg_hba, postgresql.conf, postgresql.auto.conf, alter system, and the hba, config and ident file options.

  • Approve this guard exemption

Merge after #258: both touch scripts/runner-start-test.sh; this PR syncs onto main once #258 lands.

Gives the Mac's fleet account a PostgreSQL 16 server for test suites, installed and run entirely inside that account, and stopping whenever the Mac's owner is using it.

What it does

  • Install: postgres is pinned in the offload mise file next to the other tools, built from source with no icu, readline, openssl or uuid so it needs nothing from the owner's Homebrew. No sudo, nothing in the owner's account or settings.
  • Data: a cluster in the fleet account's home (osx_postgres_data), created with initdb only when the directory has no PG_VERSION.
  • Start: the account only logs in over ssh, so there is no launchd domain. Its crontab starts the server at boot and re-checks every few minutes, through the runners' starter (runner-start.sh, now told by two variables which gate to run and what to name its pid file). A new postgres-gate.sh runs niced and does nothing if one is already running.
  • Access: peer authentication on a Unix socket in a 0700 directory in the fleet account's home, SCRAM passwords over localhost, listening on localhost only. A random password is written once to the account's ~/.pgpass (0600, never printed or logged, kept if a line exists) and set on the role through a peer connection with the statement on stdin, so no process list shows it. The play fails if a stored password is shorter than the 32 characters it generates. The fleet role is created with createuser -s only if missing, which covers any databases a suite creates.
  • Connections: max_connections is sized from the Mac's cores, a partition count (default 4) and headroom, never below 100.

How it yields to the owner

The gate starts the server only when the Mac is free by the runners' own rule: keyboard and mouse idle for the configured time and the one-minute load under the cap. When input arrives it runs pg_ctl stop -m fast: sessions are disconnected and their open transactions rolled back, committed data is flushed, and nothing is killed mid-write. A test run in flight fails and is re-run. It starts again at the next idle check. Load alone never stops it, since a test run loads the Mac through the server it is using; this differs from the runner, which lets a running job finish.

Checks

  • scripts/postgres-gate-test.sh (new) covers start, fast stop, load-only, role creation and password handling, and refused settings; runner-start-test.sh covers the new pid-file prefix; osx-offload-guard-test.sh covers the new tasks, including running the .pgpass expressions through jinja2. All pass locally, with yamllint, ansible-lint and shellcheck.
  • Not run: the playbook with --check and anything on the Mac itself (out of scope for this change; the Mac was in use). The build on the Mac, and peer authentication over the socket there, are unverified until the first run.

Docs: docs/offload-host.md, new PostgreSQL section, including how a suite connects over the socket or TCP.

CI fix: the gate doubled single quotes in the role password with a bash 4 only escape, which bash 3.2 (the Mac's /bin/bash) leaves as backslashes; it now uses a variable holding the quote. scripts/postgres-gate-test.sh passes on Linux and on the Mac under /bin/bash.

Sync with main

Merged main into this branch (a merge commit) after the Mac runner service and the git config task landed. The only conflicts were two test suites, and both kept both sides' checks: the offload guard suite now sweeps offload-postgres.yml and offload-git.yml for become and keeps the postgres and git sections, and the runner starter suite keeps main's caffeinate, work-directory and refresh checks before the postgres prefix check (which waits for the 6th logged start: main's caffeinate stub never logs one). The runner starter now wraps every gate in caffeinate -i, so the pid file names caffeinate; the docs now say to stop the postgres and redis gates with pkill -f as well, and the starter's header says so. Review: polish and panel-review (gemini) over the merge delta, PR-Agent as the bot review (cubic refused again on quota).

Step records

Run 000001

Steps at pre-implementation (run 000001)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 488246ded015.

Steps at pre-ci (run 000001)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 17db4ba91fb1.

Steps at convergence (run 000001)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 polish skill polish in-session in-session 17db4ba to 9cc482d 2026-11-12T16:47:53Z 2026-11-12T18:19:45Z applied (degraded: hosted in-session; inline walks of the diff, not the full lens fan-out) Inline walk of the whole diff (small, config and shell). Applied: a stored .pgpass password shorter than 32 characters now fails the play (finding relayed from mission control); a role or password step that failed right after the server started is retried each pass; docs describe both. Second walk found nothing new. Mac untouched; only local targeted suites ran (postgres-gate, runner-start, runner-gate, osx-offload-guard, shellcheck, yamllint, ansible-lint).
2 panel-review skill panel-review in-session in-session 9cc482d to 4f51bc4 2026-11-12T18:19:55Z 2026-11-12T18:49:11Z applied (degraded: hosted in-session; gemini backend run inline, no extra session) Gemini, two passes. Pass 1 applied: refuse unsafe role, port, connection-cap and socket-directory settings in the postgres gate and a path-like pid prefix in the starter; run the .pgpass expressions through jinja2 in the guard test; match the .pgpass write condition's regex escape to the length assert's (spelling only); correct the docs on what another account can reach. Declined: the claim that the assert's regex always fails (false: the expressions were executed and behave), cron paths with spaces, start retry without backoff. Pass 2 applied nothing: remaining items were pre-existing sibling-script patterns, role-controlled values or extra failure-case tests.

Ended on 4f51bc414934.

Steps at pre-pr (run 000001)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 4f51bc414934.

Steps at post-pr (run 000001)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 bot-review skill bot-review in-session in-session 4f51bc4 2026-11-12T18:57:51Z 2026-11-12T19:35:49Z halted (degraded: hosted in-session; no extra session dispatched) Cubic answered this PR's first trigger with a quota refusal: "cubic can't start this review because your workspace has reached its free monthly review limit. cubic has reviewed 40,041 of the 40,000 allowed lines of code this month. Reviews resume on 2 December 2026." Switched to PR-Agent (its first run timed out; one rerun succeeded on this head). Two findings, both in its summary, none posted inline, nothing replied or recorded. (1) Rejected, queued: an existing ~/.pgpass that already holds the role's line is not re-chmodded, since the write is skipped; valid, but a secrets-file zone, so queued for sign-off with the fix (enforce 0600 with a file task after the write, skipped in check mode) unapplied. (2) Rejected, draft reply queued: the two-pass sed unescape of .pgpass is correct, checked against 400 random escaped passwords with a reference decode and by the gate suite. Halted for human attention on (1).

Ended on 4f51bc414934.

Run 000002

Run 000003

Steps at pre-implementation (run 000003)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 851271d6494b.

Steps at convergence (run 000003)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 polish skill polish in-session inline 851271d to ef4102c 2026-11-13T03:23:10Z 2026-11-13T06:59:20Z applied Iteration 1: nine single-lens agents over the whole branch diff against main. Applied in bf9bfef: the gate-now task starts from the account's home, the superuser sentence in docs/offload-host.md corrected, the mise [env] comment fixed, a vacuous kill assertion dropped. Declined with reasons: gate design points the docs already state or that match the sibling runner (retry without backoff, role password applied once per server lifetime, fast stop on one idle misread, stock durability settings), hardening outside this unit (PGREQUIREAUTH, non-superuser role), items unverifiable here (initdb file ordering, bash 3.2 quoting), the Redis start task's own cwd (a separate merged unit). / Iteration 2: nine lenses in three agents over 851271d..bf9bfef. One finding applied, in ef4102c: a guard-test assertion pinning the chdir. Cosmetic column-width notes declined. / Iteration 3: nine lenses over bf9bfef..ef4102c (the one test line): nothing applied; converged.
2 panel-review skill panel-review in-session gemini ef4102c to 598916a 2026-11-13T06:59:36Z 2026-11-13T07:45:46Z applied Backend: gemini (this machine's profile default; upload already approved for this repository). Hosted in the unit session, not an isolated one. / Iteration 1 over ef4102c: one finding validated and applied in 598916a, the postgres role is now created in the same pass that starts the server (a client arriving right after the start no longer finds it missing). Declined: gate-termination trap not exercised by the suite (the stop path it calls is), shellcheck SC2030/SC2031 infos (intentional subshell exports), an ansible-lint crash in its own config dump (environment, not the change), shape guards on values the role sets (poll, dir, pgpass, cache, nice), which reach no shell, SQL or option string. / Iteration 2 over 598916a: the same declines, nothing new; converged. Both gemini runs exited 0.

Ended on 598916a0543c.

Steps at pre-pr (run 000003)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 598916a0543c.

Steps at post-pr (run 000003)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 598916a0543c.

Run 000004

Steps at post-pr (run 000004)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 bot-review skill bot-review in-session in-session 598916a 2026-11-13T07:46:14Z 2026-11-13T13:53:18Z passed cubic refused this PR for quota, in its own reply on the PR: "cubic can't start this review because your workspace has reached its free monthly review limit. cubic has reviewed 40,041 of the 40,000 allowed lines of code this month. Reviews resume on 2 December 2026", so PR-Agent is the bot review. PR-Agent run 43388129261.1 on 598916a finished with success (about seventeen minutes after the push). Its earlier finding that an existing .pgpass keeps a loose mode was fixed on the operator's ruling (mode held to 0600 on every run) and PR-Agent now lists it resolved at 598916a. Two active findings, both declined with evidence in replies on the PR and in the ledger: the .pgpass unescaping (checked against the exact sed expression, decodes correctly) and the combined port and connection-cap check (an empty value cannot reach it, the defaults use the colon-dash form). Nothing applied, nothing pushed; reviewed head equals the head.

Ended on 598916a0543c.

Run 000005

Steps at pre-implementation (run 000005)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 598916a0543c.

Steps at pre-ci (run 000005)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 1a2d76fcd6a9.

Steps at convergence (run 000005)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 polish skill polish in-session adopter 1a2d76f to 42b084d 2026-11-14T06:37:39Z 2026-11-14T06:46:14Z applied (degraded: in-session inline review of a two-line fix, not a subagent fan-out: pass one added a why-comment (42b084d); pass two found nothing; gate suite passes locally and on the Mac under bash 3.2)
2 panel-review skill panel-review in-session adopter 42b084d 2026-11-14T06:46:33Z 2026-11-14T06:55:06Z passed (degraded: ran in-session: one Gemini discovery pass, validated locally, converged with nothing applied; two low-reachability hardening notes (an unguarded path or interval variable in the gate and the runner start script, and grep without -- on an operator-set pgpass path) declined as out of scope for a test-fix and unreachable from trusted config)

Ended on 42b084d65956.

Steps at pre-pr (run 000005)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 42b084d65956.

Run 000006

Steps at pre-implementation (run 000006)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 42b084d65956.

Steps at pre-ci (run 000006)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on e4f773ce0340.

Steps at convergence (run 000006)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 polish skill polish in-session adopter e4f773c to 497f431 2026-11-14T08:10:18Z 2026-11-14T08:47:49Z applied (degraded: ran in-session through /self-review --nested with nine Sonnet lens sub-agents per pass over the delta from 42b084d, looped until a pass found no defect; passes applied the exemption narrowing to the gate's loopback value, the hyphenated option spellings and prose fixes (e4f773c to 497f431))
2 panel-review skill panel-review in-session adopter 497f431 2026-11-14T08:46:00Z 2026-11-14T08:47:49Z passed (degraded: in-session: Gemini passes alternated with the polish passes; this final one at 497f431 found only repeats of items declined earlier)

Ended on 497f4314d22d.

Steps at pre-pr (run 000006)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 497f4314d22d.

Steps at post-pr (run 000006)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 bot-review skill bot-review in-session adopter 497f431 2026-11-14T08:54:28Z 2026-11-14T09:06:38Z passed (degraded: ran /bot-review --nested --reviewer pr_agent in-session; cubic's refusal is quoted in the excerpt) cubic refused this PR on 2026-11-12T18:57:57Z: "cubic can't start this review because your workspace has reached its free monthly review limit. cubic has reviewed 40,041 of the 40,000 allowed lines of code this month. Reviews resume on 2 December 2026." Under the cubic-quota rule PR-Agent's review counts as the bot review. / PR-Agent run 43577999273.1 reviewed 497f431 (success): 0 reviews, 0 inline findings, 0 description-level findings, not errored; reviewed head equals HEAD, nothing to reply to or resolve.

Ended on 497f4314d22d.

Ended on 497f4314d22d.

Run 000007

Steps at pre-implementation (run 000007)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 6a21df5a6458.

Steps at pre-ci (run 000007)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 6a21df5a6458.

Steps at convergence (run 000007)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 polish skill polish in-session stream-json-persistent 6a21df5 to 1f2ee84 2026-11-14T16:09:31Z 2026-11-14T16:40:56Z applied (degraded: per-step session not spawned: lens fan-out ran as foreground Agent calls in the unit session) polish --nested over the merge delta: 4 passes (9-lens fan-out, 2 lens, 1 lens, 1 lens); 3 doc/comment fixes applied (stale stop recipes under the caffeinate wrapper, runner-start.sh pid-file header, one pkill pattern span); final pass applied nothing. Declined as pre-existing on main: vacuous caffeinate stub check in runner-start-test.sh, refresh pgrep/pkill window.
2 panel-review skill panel-review in-session stream-json-persistent 1f2ee84 2026-11-14T16:41:07Z 2026-11-14T16:59:37Z passed (degraded: per-step session not spawned: ran in the unit session) panel-review --nested (gemini): 1 iteration, 4 findings (low/medium) all in content that came from main and untouched by the merge: refresh.sh unescaped dots in pgrep pattern, job-started-hook :? message, tpm-guard lacking set -eu, identifier-check default-scope readability guard. All declined with rationale (pre-existing, not merge-introduced, fail closed or theoretical); nothing applied; converged.

Ended on 1f2ee84dc4cb.

Steps at pre-pr (run 000007)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
none

Ended on 1f2ee84dc4cb.

Steps at post-pr (run 000007)

# Step Kind Target Hosting Backend Session Head Start End Outcome Excerpt Output
1 bot-review skill bot-review in-session stream-json-persistent 1f2ee84 2026-11-14T16:59:37Z 2026-11-14T17:26:33Z passed (degraded: ran in the unit session) cubic triggered first on the new head; its reply on the PR refused: "cubic can't start this review because your workspace has reached its free monthly review limit ... 40,041 of the 40,000 allowed lines ... Reviews resume on 2 December 2026". Switched to PR-Agent (pr_agent): its run on the head succeeded and its review is updated to this head. Its two focus areas repeat findings already rejected with replies in the ledger (recorded-reply, nothing new); no unresolved finding; converged.

Ended on 1f2ee84dc4cb.


Summary by cubic

Runs a PostgreSQL 16 server in the Mac's fleet account so test suites have a real database, installed and run entirely inside that account and stopping fast when the Mac's owner starts using it. It starts only while the Mac is free by the runners' idle rule; load alone never stops it, since a test run loads the Mac through the server it uses.

Access

  • Peer authentication on a Unix socket in a 0700 directory; SCRAM passwords over localhost, listening on localhost only.
  • A random 32-character password is written once to ~/.pgpass and set on the role through stdin (its escapes hold on the Mac's bash 3.2); ~/.pgpass is held to 0600 on every run, and the play fails when a stored password is shorter.
  • The fleet role is created as a superuser only when missing, in the same pass the server starts, so a client arriving as it opens finds it.

Keeping it running

  • postgres is pinned in the offload mise file, built from source without icu, readline, openssl or uuid so it needs nothing from the owner's Homebrew; no sudo.
  • A crontab starts the server at boot and re-checks every few minutes through the runners' starter, which now takes RUNNER_GATE and RUNNER_PID_PREFIX; the gate keeps its pid and logs under its own prefix, runs from the account's home so the mise shim finds its config, and runs under caffeinate -i so its idle waits are not cut short by system sleep.
  • The role and password are re-checked each pass until they succeed; the server settings reach postgres as command-line options, and the no-managed-config guard exempts only that file and only listen_addresses=localhost, still rejecting pg_hba.conf, postgresql.conf, postgresql.auto.conf and ALTER SYSTEM.
  • New postgres-gate-test.sh covers start, fast stop, role creation and password handling; the runner-start, offload-guard and services-declaration suites were extended. docs/offload-host.md gained a PostgreSQL section, including how to stop the server for good.
  • The playbook was not run on the Mac itself (it was in use), so the build there and peer auth over the socket are unverified until the first run.

Written for commit 1f2ee84. Summary will update on new commits.

View guided diff Turn on auto-fix

… its owner

mise installs the server in the fleet account. Its crontab starts it through the runners' starter, and a gate starts it only while the Mac is idle and stops it fast when the owner returns. Peer on a private socket, scram over localhost, password in the account's .pgpass.
A .pgpass line kept from a hand edit or an earlier task version could carry a short or empty password that the gate then applies on every start. The play now asserts the stored password is at least as long as the generated one.
A role or password step that failed right after the server started was never tried again until the next restart. The gate now repeats it on each pass until it succeeds once.
@inkatze
inkatze commented Nov 12, 2026
Copy link
Copy Markdown
Owner Author

@cubic-dev-ai review this PR

@cubic-dev-ai
cubic-dev-ai Bot commented Nov 12, 2026
Copy link
Copy Markdown

@cubic-dev-ai review this PR

@inkatze cubic can't start this review because your workspace has reached its free monthly review limit. cubic has reviewed 40,041 of the 40,000 allowed lines of code this month. Reviews resume on 2 December 2026 (in 21 days). Paid plans include much higher monthly review limits. Upgrade now to resume reviews.

To help optimise your usage, you can tune cubic to get the most out of your usage limits:

Learn more →

@github-actions
github-actions Bot commented Nov 12, 2026 •
Copy link
Copy Markdown

PR-Agent run 43618666473.1 on 1f2ee84: success

@github-actions
github-actions Bot commented Nov 12, 2026 •
Copy link
Copy Markdown

PR Reviewer Guide 🔍

(Review updated until commit 1f2ee84)

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 3 🔵🔵🔵⚪⚪
🧪 PR contains tests
🔒 No security concerns identified
⚡ Recommended focus areas for review

Incomplete port and max_connections validation

The combined case check on "$port$max_connections" accepts an empty POSTGRES_PORT or POSTGRES_MAX_CONNECTIONS as long as the other value is all digits, because concatenating an empty string with digits yields an all-digit string. This lets the gate reach pg_ctl with -c port= or -c max_connections=, which PostgreSQL rejects and the server fails to start. Validate port and max_connections separately against '' | [!0-9].

case "$port$max_connections" in '' | *[!0-9]*) echo "postgres-gate: port and connection cap must be numbers" >&2; exit 1 ;; esac
Incorrect .pgpass unescaping

pgpass_password reverses pgpass escaping with two sed passes: first \ -> , then : -> :. A password that literally contains a backslash followed by a colon is encoded in .pgpass as three backslashes and a colon (e.g. \:). The first pass collapses the doubled backslash before the colon escape is handled, so the decoded password is corrupted. The gate then sets the role password to a different string than the one stored in ~/.pgpass, and psql reading .pgpass fails to authenticate over TCP.

pgpass_password() {
    local line
    line="$(grep -m1 "^localhost:$port:\*:$role:" "$pgpass" 2>/dev/null)" || return 0
    printf '%s\n' "${line#localhost:"$port":\*:"$role":}" | sed -e 's/\\\\/\\/g' -e 's/\\:/:/g'
}
✅ Resolved findings

roles/osx/files/runner/postgres-gate.sh:42-46

Incorrect .pgpass unescaping

pgpass_password reverses pgpass escaping with two sed passes: first \ -> , then \: -> :. A password that literally contains a backslash followed by a colon is encoded in .pgpass as three backslashes and a colon (e.g. \\:)....

roles/osx/files/runner/postgres-gate.sh:42-46

Incorrect .pgpass unescaping

pgpass_password reverses pgpass escaping with two sed passes: first \\ -> \, then \: -> :. A password that literally contains a backslash followed by a colon is encoded in .pgpass as three backslashes and a colon (e.g. \:). The first pass collapses the doubled backslash before the colon escape is handled, so the decoded password is corrupted. The gate then sets the role password to a different string than the one stored in ~/.pgpass, and psql reading .pgpass fails to authenticate over TCP.

roles/osx/tasks/offload-postgres.yml:60-72

Insecure ~/.pgpass permissions

The lineinfile task that writes ~/.pgpass sets mode 0600 only when it creates the file. If a .pgpass already exists with looser permissions (for example from a hand edit or an earlier setup), those permissions are left unchanged, leaving the generated PostgreSQL password readable by group/other. Add a task that enforces 0600 on every run.

@github-actions
Copy link
Copy Markdown

Persistent review updated to latest commit 598916a

@inkatze
inkatze commented Nov 13, 2026
Copy link
Copy Markdown
Owner Author

Declined: the two-pass sed in pgpass_password decodes libpq escaping correctly, because \\ is turned into \ first and only then does \: become :. Checked with the exact expression: the stored form of the password a\:b (a\\\:b) decodes to a\:b, and the stored form of \\: (\\\\\:) decodes to \\:. The guard test also round-trips a password holding an escaped colon, an escaped backslash and a quote. The generated password is 32 alphanumeric characters, so none of these characters occur in it today.

@inkatze
inkatze commented Nov 13, 2026
Copy link
Copy Markdown
Owner Author

Declined: an empty value cannot reach this check. port and max_connections are read as ${POSTGRES_PORT:-5432} and ${POSTGRES_MAX_CONNECTIONS:-100}, and :- substitutes the default for an empty value as well as an unset one, so each is always non-empty before the digit test and the concatenation hides no case. A non-numeric value of either is still refused by the same test, and the guard suite covers it.

@github-actions
Copy link
Copy Markdown

Persistent review updated to latest commit 42b084d

…ses as command-line options

The gate starts PostgreSQL with these settings as options and writes no configuration file, so the no-managed-config guard exempts that one file.
…e config files

A later write to pg_hba.conf or postgresql.conf from the gate would still fail the guard; the header and scope comments now name the exception.
… other config route

The exemption now allows only that value and still rejects pg_hba, postgresql.conf and postgresql.auto.conf, ALTER SYSTEM, hba_file and config_file; widening the bound address fails the guard.
@github-actions
Copy link
Copy Markdown

Persistent review updated to latest commit 497f431

Main added the Mac runner service and the offload git config; both sides
changed the osx offload guard suite and the runner starter suite, so each
conflict keeps both sides' checks.

- osx-offload-guard-test.sh: the no-sudo sweep covers offload-postgres.yml and
  offload-git.yml; the postgres and git sections both stay.
- runner-start-test.sh: main's caffeinate, work-directory and refresh checks
  come first, then the postgres prefix check. The prefix check waits for start
  6, not 5: main's caffeinate stub execs "-i", so that start never logs, and
  the count stands at 5 when the postgres gate starts as the 6th.
…feinate above them

The shared starter now runs every gate under caffeinate -i, so the pid file
holds caffeinate, not the gate. The runner section already said to stop the
gate too; the postgres and redis sections still told the operator to kill the
pid file alone, which leaves the gate and its server running. Both now say so
and note the gate holds off idle system sleep.
…when present

The wrapper added to the starter means the pid file holds caffeinate, not the
gate; the header still said it names the gate, inviting an edit to running()
or a kill of the wrong process.
A pattern wrapped across a source line copies with a newline inside the quotes,
so pkill matches nothing and says so silently.
@inkatze
inkatze commented Nov 14, 2026
Copy link
Copy Markdown
Owner Author

@cubic-dev-ai review this PR

@cubic-dev-ai
cubic-dev-ai Bot commented Nov 14, 2026
Copy link
Copy Markdown

@cubic-dev-ai review this PR

@inkatze cubic can't start this review because your workspace has reached its free monthly review limit. cubic has reviewed 40,041 of the 40,000 allowed lines of code this month. Reviews resume on 2 December 2026 (in 19 days). Paid plans include much higher monthly review limits. Upgrade now to resume reviews.

To help optimise your usage, you can tune cubic to get the most out of your usage limits:

Learn more →

@github-actions
Copy link
Copy Markdown

Persistent review updated to latest commit 1f2ee84

@inkatze
inkatze marked this pull request as ready for review November 14, 2026 17:55
@inkatze
inkatze merged commit a23f611 into main Nov 14, 2026
14 checks passed
@inkatze
inkatze deleted the planwright/flight/mc-dotfiles-ad-hoc-3117da64 branch November 14, 2026 19:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant

0