mirror of
https://github.com/OpenRouterTeam/spawn.git
synced 2026-05-20 01:11:18 +00:00
fix(e2e): improve openclaw reliability on AWS and other clouds (#2123)
* fix(e2e): improve openclaw reliability on AWS and other clouds Three changes to make openclaw e2e tests more robust: 1. Increase PROVISION_TIMEOUT from 480s to 720s — AWS cloud-init for "full" tier (Node.js + Bun + build-essential) can exceed 480s, causing the CLI to be killed before .spawnrc is written. 2. Add .spawnrc manual fallback in provision.sh — if the CLI is killed before writing .spawnrc, construct it via SSH using OPENROUTER_API_KEY with agent-specific env vars (openclaw, zeroclaw). 3. Add retry logic to openclaw gateway input test — the gateway can crash with 1006 websocket closure on resource-constrained instances. Now retries once after killing and restarting the gateway process. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(security): fix command injection in e2e provision scripts - Use printf %q and temp file for api_key handling in provision.sh to prevent shell metachar injection (single quotes, backticks, $) - Double-quote env_b64 interpolation in cloud_exec call to prevent word splitting - Replace echo with printf in bashrc append to avoid portability issues - Replace overbroad pkill -f 'openclaw gateway' in verify.sh with PID-targeted kill via lsof/fuser on port 18789 Agent: pr-maintainer Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: B <6723574+louisgv@users.noreply.github.com>
This commit is contained in:
Ahmed Abushagur
GitHub
parent
7b650a0103
commit
4a90abdaa2
3 changed files with 114 additions and 25 deletions
|
|
@ -6,7 +6,7 @@ set -eo pipefail
|
|||
# Constants
|
||||
# ---------------------------------------------------------------------------
|
||||
ALL_AGENTS="claude openclaw zeroclaw codex opencode kilocode hermes"
|
||||
PROVISION_TIMEOUT="${PROVISION_TIMEOUT:-480}"
|
||||
PROVISION_TIMEOUT="${PROVISION_TIMEOUT:-720}"
|
||||
INSTALL_WAIT="${INSTALL_WAIT:-600}"
|
||||
INPUT_TEST_TIMEOUT="${INPUT_TEST_TIMEOUT:-120}"
|
||||
|
||||
|
|
|
|||
|
|
@ -127,8 +127,57 @@ CLOUD_ENV
|
|||
sleep 5
|
||||
log_ok "Install completed (.spawnrc found)"
|
||||
return 0
|
||||
else
|
||||
log_warn ".spawnrc not found after ${effective_install_wait}s — install may still be running"
|
||||
return 0 # Continue to verification; it will catch real failures
|
||||
fi
|
||||
|
||||
# Fallback: CLI was killed before writing .spawnrc (provision timeout race).
|
||||
# Construct .spawnrc manually via SSH using available env vars.
|
||||
log_warn ".spawnrc not found after ${effective_install_wait}s — attempting manual creation"
|
||||
local api_key="${OPENROUTER_API_KEY:-}"
|
||||
if [ -z "${api_key}" ]; then
|
||||
log_err "Cannot create .spawnrc fallback — OPENROUTER_API_KEY not set"
|
||||
return 0
|
||||
fi
|
||||
|
||||
# Build env lines in a temp file to avoid interpolating api_key into shell
|
||||
# strings directly (prevents command injection if the key contains shell
|
||||
# metacharacters like single quotes, backticks, or dollar signs).
|
||||
local env_tmp
|
||||
env_tmp=$(mktemp)
|
||||
{
|
||||
printf '%s\n' "# [spawn:env]"
|
||||
printf 'export IS_SANDBOX=%q\n' "1"
|
||||
printf 'export OPENROUTER_API_KEY=%q\n' "${api_key}"
|
||||
} > "${env_tmp}"
|
||||
|
||||
# Add agent-specific env vars
|
||||
case "${agent}" in
|
||||
openclaw)
|
||||
{
|
||||
printf 'export ANTHROPIC_API_KEY=%q\n' "${api_key}"
|
||||
printf 'export ANTHROPIC_BASE_URL=%q\n' "https://openrouter.ai/api"
|
||||
} >> "${env_tmp}"
|
||||
;;
|
||||
zeroclaw)
|
||||
{
|
||||
printf 'export ZEROCLAW_PROVIDER=%q\n' "openrouter"
|
||||
printf 'export OPENAI_API_KEY=%q\n' "${api_key}"
|
||||
printf 'export OPENAI_BASE_URL=%q\n' "https://openrouter.ai/api/v1"
|
||||
} >> "${env_tmp}"
|
||||
;;
|
||||
esac
|
||||
|
||||
local env_b64
|
||||
env_b64=$(base64 < "${env_tmp}" | tr -d '\n')
|
||||
rm -f "${env_tmp}"
|
||||
|
||||
# Use double-quoting around env_b64 in the remote command to prevent word
|
||||
# splitting. Base64 output is shell-safe ([A-Za-z0-9+/=]), but quoting is
|
||||
# defensive best practice against any upstream corruption.
|
||||
if cloud_exec "${app_name}" "printf '%s' \"${env_b64}\" | base64 -d > ~/.spawnrc && chmod 600 ~/.spawnrc && \
|
||||
grep -q 'source ~/.spawnrc' ~/.bashrc 2>/dev/null || printf '%s\n' '[ -f ~/.spawnrc ] && source ~/.spawnrc' >> ~/.bashrc" >/dev/null 2>&1; then
|
||||
log_ok "Manual .spawnrc created successfully"
|
||||
else
|
||||
log_err "Failed to create manual .spawnrc"
|
||||
fi
|
||||
return 0
|
||||
}
|
||||
|
|
|
|||
|
|
@ -74,12 +74,8 @@ input_test_codex() {
|
|||
fi
|
||||
}
|
||||
|
||||
input_test_openclaw() {
|
||||
_openclaw_ensure_gateway() {
|
||||
local app="$1"
|
||||
|
||||
log_step "Running input test for openclaw..."
|
||||
|
||||
# Ensure the gateway is running (it may have died after provisioning)
|
||||
log_step "Ensuring openclaw gateway is running on :18789..."
|
||||
cloud_exec "${app}" "source ~/.spawnrc 2>/dev/null; \
|
||||
export PATH=\$HOME/.npm-global/bin:\$HOME/.bun/bin:\$HOME/.local/bin:\$PATH; \
|
||||
|
|
@ -99,27 +95,71 @@ input_test_openclaw() {
|
|||
log_err "OpenClaw gateway failed to start"
|
||||
return 1
|
||||
fi
|
||||
}
|
||||
|
||||
_openclaw_restart_gateway() {
|
||||
local app="$1"
|
||||
log_step "Restarting openclaw gateway..."
|
||||
cloud_exec "${app}" "source ~/.spawnrc 2>/dev/null; \
|
||||
export PATH=\$HOME/.npm-global/bin:\$HOME/.bun/bin:\$HOME/.local/bin:\$PATH; \
|
||||
_gw_pid=\$(lsof -ti tcp:18789 2>/dev/null || fuser 18789/tcp 2>/dev/null | tr -d ' ') && \
|
||||
kill \"\$_gw_pid\" 2>/dev/null; sleep 2; \
|
||||
_oc_bin=\$(command -v openclaw) || exit 1; \
|
||||
if command -v setsid >/dev/null 2>&1; then setsid \"\$_oc_bin\" gateway > /tmp/openclaw-gateway.log 2>&1 < /dev/null & \
|
||||
else nohup \"\$_oc_bin\" gateway > /tmp/openclaw-gateway.log 2>&1 < /dev/null & fi; \
|
||||
elapsed=0; while [ \$elapsed -lt 30 ]; do \
|
||||
if (echo >/dev/tcp/127.0.0.1/18789) 2>/dev/null || nc -z 127.0.0.1 18789 2>/dev/null; then echo 'Gateway restarted'; break; fi; \
|
||||
sleep 1; elapsed=\$((elapsed + 1)); \
|
||||
done" >/dev/null 2>&1 || log_warn "Failed to restart openclaw gateway"
|
||||
}
|
||||
|
||||
input_test_openclaw() {
|
||||
local app="$1"
|
||||
local max_attempts=2
|
||||
local attempt=0
|
||||
|
||||
log_step "Running input test for openclaw..."
|
||||
|
||||
local encoded_prompt
|
||||
encoded_prompt=$(printf '%s' "${INPUT_TEST_PROMPT}" | base64 -w 0 2>/dev/null || printf '%s' "${INPUT_TEST_PROMPT}" | base64)
|
||||
local remote_cmd
|
||||
remote_cmd="source ~/.spawnrc 2>/dev/null; \
|
||||
export PATH=\$HOME/.npm-global/bin:\$HOME/.bun/bin:\$HOME/.local/bin:\$PATH; \
|
||||
rm -rf /tmp/e2e-test && mkdir -p /tmp/e2e-test && cd /tmp/e2e-test && git init -q; \
|
||||
PROMPT=\$(printf '%s' '${encoded_prompt}' | base64 -d); openclaw agent --message \"\$PROMPT\" --session-id e2e-test --json"
|
||||
|
||||
local output
|
||||
output=$(cloud_exec_long "${app}" "${remote_cmd}" "${INPUT_TEST_TIMEOUT}" 2>&1) || true
|
||||
while [ "${attempt}" -lt "${max_attempts}" ]; do
|
||||
attempt=$((attempt + 1))
|
||||
|
||||
if printf '%s' "${output}" | grep -q "${INPUT_TEST_MARKER}"; then
|
||||
log_ok "openclaw input test — marker found in response"
|
||||
return 0
|
||||
else
|
||||
log_err "openclaw input test — marker '${INPUT_TEST_MARKER}' not found in response"
|
||||
log_err "Response (last 5 lines):"
|
||||
printf '%s\n' "${output}" | tail -5 >&2
|
||||
return 1
|
||||
fi
|
||||
# Ensure/restart gateway
|
||||
if [ "${attempt}" -eq 1 ]; then
|
||||
_openclaw_ensure_gateway "${app}"
|
||||
else
|
||||
log_warn "Retrying openclaw input test (attempt ${attempt}/${max_attempts})..."
|
||||
_openclaw_restart_gateway "${app}"
|
||||
fi
|
||||
|
||||
local remote_cmd
|
||||
remote_cmd="source ~/.spawnrc 2>/dev/null; \
|
||||
export PATH=\$HOME/.npm-global/bin:\$HOME/.bun/bin:\$HOME/.local/bin:\$PATH; \
|
||||
rm -rf /tmp/e2e-test && mkdir -p /tmp/e2e-test && cd /tmp/e2e-test && git init -q; \
|
||||
PROMPT=\$(printf '%s' '${encoded_prompt}' | base64 -d); openclaw agent --message \"\$PROMPT\" --session-id e2e-test-${attempt} --json --timeout 60"
|
||||
|
||||
local output
|
||||
output=$(cloud_exec_long "${app}" "${remote_cmd}" "${INPUT_TEST_TIMEOUT}" 2>&1) || true
|
||||
|
||||
if printf '%s' "${output}" | grep -q "${INPUT_TEST_MARKER}"; then
|
||||
log_ok "openclaw input test — marker found in response"
|
||||
return 0
|
||||
fi
|
||||
|
||||
if [ "${attempt}" -lt "${max_attempts}" ]; then
|
||||
log_warn "openclaw input test attempt ${attempt} failed — will retry"
|
||||
log_warn "Response (last 3 lines):"
|
||||
printf '%s\n' "${output}" | tail -3 >&2
|
||||
else
|
||||
log_err "openclaw input test — marker '${INPUT_TEST_MARKER}' not found in response"
|
||||
log_err "Response (last 5 lines):"
|
||||
printf '%s\n' "${output}" | tail -5 >&2
|
||||
fi
|
||||
done
|
||||
|
||||
return 1
|
||||
}
|
||||
|
||||
input_test_zeroclaw() {
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue