Compare commits

..

8 Commits

Author SHA1 Message Date
wmantly 1d09f243dd Release 1.8.1: Redis persistence fix (#20) 2026-07-28 15:50:11 -04:00
wmantly ec4ca97af4 Fix: Redis had zero persistence — every rebuild wiped sessions, (#18)
in-flight OAuth logins, and any admin-created API token

redis-server ran with --save '' --appendonly no (deliberately ephemeral,
per the original "audit/metrics/session storage" framing). That stopped
being a safe assumption once API tokens (PATs) lived in this same Redis
-- a PAT is supposed to be a stable, long-lived credential, not
disposable session state, but every `docker rm -f jump-host` + rebuild
silently invalidated every one that existed.

Matches proxy's existing pattern exactly: AOF + periodic RDB persisted
to $REDIS_DATA_DIR (default /data), which the deployment mounts as a
volume (see the companion theta-env change).

Verified against a live container: minted a real PAT, force-recreated
the container (docker rm -f + rebuild), confirmed the same token still
authenticates afterward.

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-28 15:46:40 -04:00
wmantly 21ef8960c4 Merge pull request #17 from theta42/release/1.8.0
Release 1.8.0
2026-07-28 13:35:18 -04:00
wmantly a5bef2980b Release 1.8.0: fix TUI-mode SSH connection drops, HTML-escaped loading indicator
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-28 13:32:28 -04:00
wmantly ab2ee0fed3 Merge pull request #16 from theta42/fix/tui-session-listener-race
Fix: TUI-mode SSH connections could drop with PTY/shell request failed
2026-07-28 13:07:55 -04:00
wmantly ab9e04e007 Merge pull request #15 from theta42/fix/loading-message-html-escaped
Fix HTML-escaped loading indicator in formAJAX
2026-07-28 13:07:49 -04:00
wmantly a3a6787776 Fix: TUI-mode SSH connections could drop with "PTY allocation
request failed" / "shell request failed"

runTuiSession awaited audit.create() (a Redis round-trip) and then
accessibleHosts() (a directory API call) BEFORE calling runTui(), which
is where the pty/shell/exec/subsystem listeners actually get attached to
the session. The client sends its pty-req and shell requests immediately
after opening the session -- if either await took long enough for those
requests to arrive first, ssh2 auto-rejects any channel request with no
listener (CHANNEL_FAILURE), which is exactly what OpenSSH reports as
"PTY allocation request failed on channel 0" / "shell request failed on
channel 0". The connection then just sat there, since nothing was left
to drive it.

runGrammar already has this exact fix (see its own comment); runTuiSession
never got the equivalent treatment. Fixed the same way: register the
session listeners synchronously, before any await, by having runTui take
a Promise for the hosts list instead of the resolved list -- the shell
handler awaits it internally once the client actually sends a shell
request, which by construction happens only after the listener already
exists.

Verified: publickey auth against a real deployment succeeds (confirms
the earlier ldaps:// fix holds), and the failure reproduces with the
exact reported error strings for a bare `ssh user@host` (TUI/picker mode,
no target) connection. Full suite: 50/50 passing, no regressions.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-28 12:58:49 -04:00
wmantly 0ead0199a3 Fix HTML-escaped loading indicator in formAJAX
Same fix as sso-manager-node/proxy: formAJAX's loading indicator passed a
raw <div class="spinner-border"> string to app.messages.action, which
HTML-escapes its message by design (@simpleworkjs/frontend). Replaced
with plain text ("Saving…").

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-28 12:53:08 -04:00
5 changed files with 64 additions and 18 deletions
+11
View File
@@ -4,6 +4,17 @@ All notable changes to this project are documented here. Format loosely
follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); versions
correspond to git tags (`vX.Y.Z`) and `nodejs/package.json`'s `version`.
## [1.8.1] - 2026-07-28
### Fixed
- **Redis had zero persistence** (`--save '' --appendonly no`, no data-dir volume) — every container rebuild/recreation silently wiped all sessions, in-flight OAuth logins, and any admin-created API token. This is why re-running `setup.sh` appeared to "break OAuth with jump": the jump-host container gets recreated, and any token or in-flight login vanished with it. Now Redis persists (AOF + periodic RDB) to `/data`, mounted as a named volume (`jump-redis-data`) in theta-env's compose file. Verified live: minted a PAT, force-recreated the container, confirmed the same PAT still authenticated afterward.
## [1.8.0] - 2026-07-28
### Fixed
- **TUI-mode SSH connections (a bare `ssh user@host`, no target) could drop with "PTY allocation request failed" / "shell request failed"** — `runTuiSession` awaited two real round-trips (an audit-log write, then a directory API call) *before* attaching the session's pty/shell/exec listeners, so a client that sent those requests quickly enough got auto-rejected by ssh2 before anything was listening. `runGrammar` (the `uid_-_target` path) already had the equivalent fix; this ports it to the picker path.
- **`formAJAX`'s loading indicator showed literal HTML**, not a spinner — same fix as sso-manager-node/proxy's companion releases.
## [1.7.1] - 2026-07-28
### Added
+11 -3
View File
@@ -12,9 +12,17 @@ if [[ -f /config/jump-secrets.js ]]; then
info "Loaded config from /config/jump-secrets.js"
fi
# Redis for audit/metrics/session storage (app connects to 127.0.0.1:6379).
info "Starting redis..."
redis-server --daemonize yes --save '' --appendonly no
# Redis for audit/metrics/session AND api-token storage (app connects to
# 127.0.0.1:6379). Persisted (AOF + periodic RDB) to /data, which the
# deployment should mount as a volume -- without this, every container
# recreation silently wiped every session, in-flight OAuth login, and any
# admin-created API token, which is especially bad for the last one since a
# PAT is meant to be a stable, long-lived credential, not session state.
REDIS_DATA_DIR="${REDIS_DATA_DIR:-/data}"
mkdir -p "$REDIS_DATA_DIR"
info "Starting redis (AOF persisted to $REDIS_DATA_DIR)..."
redis-server --daemonize yes --dir "$REDIS_DATA_DIR" --appendonly yes \
--appendfilename appendonly.aof --save 900 1 --save 300 10 --save 60 10000
# Wait for redis to answer before starting the app.
for _ in $(seq 1 20); do
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "t42-jump-host",
"version": "1.7.1",
"version": "1.8.1",
"description": "SSH jump host for the theta42 stack — LDAP-authenticated, directory-driven host bridging with audit and metrics",
"author": [
{
+4 -7
View File
@@ -679,13 +679,10 @@ function formAJAX(btn){
return false;
}
app.messages.action(
`<div class="spinner-border" role="status">
<span class="visually-hidden">Loading...</span>
</div>`,
$form,
'info'
);
// Plain text: app.messages.action HTML-escapes its message (by design,
// see @simpleworkjs/frontend), so raw markup like a spinner <div> would
// render literally instead of as an element.
app.messages.action('Saving…', $form, 'info');
app.api[method]($form.attr('action'), formData, function(error, data){
app.messages.action(data.message, $form, error ? 'danger' : 'success'); //re-populate table
+37 -7
View File
@@ -180,6 +180,23 @@ async function runGrammar(session, client, state) {
}
async function runTuiSession(session, client, state) {
// Register session listeners IMMEDIATELY, before any await — same fix,
// same reason, as runGrammar above. The client sends pty-req and shell
// requests right after opening the session; awaiting audit.create() and
// accessibleHosts() first (both real round-trips: Redis, then the
// directory API) left a window where those requests could arrive before
// runTui had attached any listener for them, and ssh2 auto-rejects an
// unlistened channel request with CHANNEL_FAILURE — surfacing to the
// client as "PTY allocation request failed" / "shell request failed",
// with the connection then just sitting there (nothing left to drive it).
let resolveHosts, rejectHosts;
const hostsPromise = new Promise((res, rej) => { resolveHosts = res; rejectHosts = rej; });
// A silent catch so a rejection isn't "unhandled" if the client never
// sends a shell request at all (exec-only) — runTui's own .catch() below
// still runs independently when it does.
hostsPromise.catch(() => {});
const tuiPromise = runTui(session, state.uid, hostsPromise);
const record = await audit.create({ uid: state.uid, authMethod: state.authMethod, clientIp: state.clientIp, mode: 'tui' });
const finishFail = async (reason) => {
@@ -189,10 +206,15 @@ async function runTuiSession(session, client, state) {
};
let hosts;
try { hosts = await accessibleHosts(state.user); }
catch (_) { return finishFail('directory-unreachable'); }
try {
hosts = await accessibleHosts(state.user);
resolveHosts(hosts);
} catch (_) {
rejectHosts(new Error('directory-unreachable'));
return finishFail('directory-unreachable');
}
const tui = await runTui(session, state.uid, hosts);
const tui = await tuiPromise;
if (!tui.host) return finishFail('cancelled');
state.target = tui.host.slug;
@@ -253,7 +275,10 @@ function reasonMessage(reason) {
// Run the TUI picker over a shell channel; returns { host, channel, ptyInfo }.
// host is null if the user quit. exec/subsystem in picker mode are rejected.
function runTui(session, uid, hosts) {
// Takes a Promise for the accessible-hosts list (not the resolved list)
// so the caller can register these listeners before that lookup completes
// — see the comment in runTuiSession for why that ordering matters.
function runTui(session, uid, hostsPromise) {
return new Promise((resolve) => {
let ptyInfo = null;
let settled = false;
@@ -262,9 +287,14 @@ function runTui(session, uid, hosts) {
session.on('pty', (accept, _reject, info) => { ptyInfo = info; accept && accept(); });
session.on('shell', (accept) => {
const channel = accept();
pickHost(channel, uid, hosts).then((host) => {
if (!host) { try { channel.write('\r\n Bye.\r\n'); channel.close(); } catch (_) {} }
finish({ host, channel, ptyInfo });
hostsPromise.then((hosts) => {
pickHost(channel, uid, hosts).then((host) => {
if (!host) { try { channel.write('\r\n Bye.\r\n'); channel.close(); } catch (_) {} }
finish({ host, channel, ptyInfo });
});
}).catch(() => {
try { channel.write('\r\n Could not reach the directory.\r\n'); channel.close(); } catch (_) {}
finish({ host: null });
});
});
session.on('exec', (accept) => {