wangbo
56a5176c92
原因：线上 P24 验收显示 Worker 在每轮准入前逐条恢复媒体、重算候选和查询 attempts，跨地域查询导致 48 个执行槽长期只能使用约 8 至 16 个。\n\n影响：任务首次排队时保存鉴权后的 admission scope 快照，Worker 单次批量读取并只刷新动态总容量；显式重新选路保留完整慢路径。验收模拟器改到非数据库专用 Worker 节点，运行期 85% 作为立即中止硬门禁，80% 继续作为容量认证目标。\n\n验证：Go 全量测试、go vet、govulncheck、真实 PostgreSQL 迁移与跨 Store 集成测试、ShellCheck、发布及验收脚本、OpenAPI、前端 lint/test/build、pnpm audit high、Compose 与 Kubernetes 渲染均通过。
156 lines
5.7 KiB
Bash
Executable File
156 lines
5.7 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# shellcheck disable=SC2034,SC2329 # Symbols are consumed by dynamically sourced functions.
|
|
set -euo pipefail
|
|
|
|
root=$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)
|
|
tmp=$(mktemp -d)
|
|
trap 'rm -rf "$tmp"' EXIT
|
|
|
|
script=$root/scripts/cluster/run-production-acceptance.sh
|
|
|
|
# shellcheck source=scripts/cluster/common.sh
|
|
source "$root/scripts/cluster/common.sh"
|
|
ssh() {
|
|
printf '%s\n' "$@" >"$tmp/ssh-arguments"
|
|
}
|
|
cluster_ssh_key=$tmp/test-key
|
|
cluster_ssh_control_dir=$tmp/ssh-control
|
|
cluster_ssh root@cluster.invalid true
|
|
grep -Fxq 'ConnectionAttempts=3' "$tmp/ssh-arguments"
|
|
grep -Fxq 'ControlMaster=auto' "$tmp/ssh-arguments"
|
|
grep -Fxq 'ControlPersist=120' "$tmp/ssh-arguments"
|
|
grep -Fxq "ControlPath=$cluster_ssh_control_dir/%C" "$tmp/ssh-arguments"
|
|
[[ $(stat -f '%Lp' "$cluster_ssh_control_dir") == 700 ]]
|
|
|
|
awk '
|
|
/^is_release_ancestor\(\)/ { capture = 1 }
|
|
capture && /^create_and_activate_run\(\)/ { exit }
|
|
capture { print }
|
|
' "$script" >"$tmp/is-release-ancestor.sh"
|
|
|
|
# shellcheck source=/dev/null
|
|
source "$tmp/is-release-ancestor.sh"
|
|
|
|
awk '
|
|
/^workload_metrics_for_site\(\)/ { capture = 1 }
|
|
capture && /^current_max_pod_memory_mib\(\)/ { exit }
|
|
capture { print }
|
|
' "$script" >"$tmp/pressure-monitor.sh"
|
|
|
|
# shellcheck source=/dev/null
|
|
source "$tmp/pressure-monitor.sh"
|
|
|
|
cluster_root=$tmp/repository
|
|
git init -q "$cluster_root"
|
|
git -C "$cluster_root" config user.name 'Acceptance Script Test'
|
|
git -C "$cluster_root" config user.email 'acceptance-script-test@example.invalid'
|
|
git -C "$cluster_root" commit --allow-empty -qm first
|
|
first=$(git -C "$cluster_root" rev-parse HEAD)
|
|
git -C "$cluster_root" commit --allow-empty -qm second
|
|
git -C "$cluster_root" commit --allow-empty -qm third
|
|
third=$(git -C "$cluster_root" rev-parse HEAD)
|
|
|
|
is_release_ancestor "$first" "$third"
|
|
if is_release_ancestor "$third" "$first"; then
|
|
echo 'descendant release was accepted as an ancestor' >&2
|
|
exit 1
|
|
fi
|
|
if is_release_ancestor invalid "$third"; then
|
|
echo 'invalid release identifier was accepted as an ancestor' >&2
|
|
exit 1
|
|
fi
|
|
|
|
remote_kubectl() {
|
|
if [[ $* == *'get pods'* ]]; then
|
|
printf '{"items":[]}'
|
|
else
|
|
printf 0
|
|
fi
|
|
}
|
|
|
|
cluster_ssh() {
|
|
echo 'zero-replica Worker attempted to collect metrics' >&2
|
|
return 1
|
|
}
|
|
|
|
namespace=easyai
|
|
if workload_metrics_for_site worker ningbo; then
|
|
echo 'zero-replica Worker site returned metrics success' >&2
|
|
exit 1
|
|
else
|
|
metrics_status=$?
|
|
fi
|
|
[[ $metrics_status == 2 ]]
|
|
|
|
remote_kubectl() {
|
|
printf '%s\n' '{"items":[
|
|
{"metadata":{},"spec":{"nodeName":"easyai-hongkong"},"status":{"podIP":"10.42.3.10","conditions":[{"type":"Ready","status":"True"}]}},
|
|
{"metadata":{},"spec":{"nodeName":"easyai-hongkong-worker-2"},"status":{"podIP":"10.42.5.10","conditions":[{"type":"Ready","status":"True"}]}}
|
|
]}'
|
|
}
|
|
|
|
cluster_ssh() {
|
|
while IFS= read -r _; do :; done
|
|
printf '# target_host %s\n' "$1"
|
|
printf 'test_metric 1\n'
|
|
}
|
|
|
|
CLUSTER_NINGBO_HOST=root@ningbo.invalid
|
|
CLUSTER_HONGKONG_HOST=root@hongkong.invalid
|
|
CLUSTER_LOS_ANGELES_HOST=root@losangeles.invalid
|
|
CLUSTER_WORKER_NODE_4_NAME=easyai-hongkong-worker-2
|
|
CLUSTER_WORKER_NODE_4_HOST=root@hongkong-worker-2.invalid
|
|
metrics=$(workload_metrics_for_site worker hongkong)
|
|
grep -Fq '# target_host root@hongkong.invalid' <<<"$metrics"
|
|
grep -Fq '# target_host root@hongkong-worker-2.invalid' <<<"$metrics"
|
|
[[ $(grep -c '^test_metric 1$' <<<"$metrics") == 2 ]]
|
|
[[ $(metric_min test_metric <<<"$metrics") == 1 ]]
|
|
[[ $(metric_max test_metric <<<"$metrics") == 1 ]]
|
|
[[ $(metric_sum test_metric <<<"$metrics") == 2 ]]
|
|
|
|
passing_pressure_row='2026-08-01T00:00:00Z,0,0,899,149,200,84,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100'
|
|
pressure_row_passes_live_gates "$passing_pressure_row"
|
|
[[ $(pressure_row_live_gate_reason '2026-08-01T00:00:00Z,0,0,900,149,200,79,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100') == \
|
|
'oldest_wait_seconds=900 limit=900' ]]
|
|
[[ $(pressure_row_live_gate_reason '2026-08-01T00:00:00Z,0,0,899,149,200,85,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100') == \
|
|
'node_memory_percent=85 hard_limit=85' ]]
|
|
|
|
for failed_pressure_row in \
|
|
'2026-08-01T00:00:00Z,0,0,901,149,200,79,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,150,200,79,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,80,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,79,1536,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,79,1535,0,32,32,0,0,1,0,2,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,79,1535,0,32,32,0,0,0,1,2,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,79,1535,0,32,32,0,0,0,0,0,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,79,1535,0,32,32,0,0,0,0,3,48,0,8,8,0,0,0,16,16,0,0,1535,100' \
|
|
'2026-08-01T00:00:00Z,0,0,900,149,200,79,1535,0,32,32,0,0,0,0,2,48,0,8,8,0,0,0,16,16,0,0,1536,100'; do
|
|
if pressure_row_passes_live_gates "$failed_pressure_row"; then
|
|
echo "unsafe pressure row passed live gates: $failed_pressure_row" >&2
|
|
exit 1
|
|
fi
|
|
done
|
|
|
|
long_running_load() {
|
|
while :; do
|
|
sleep 1
|
|
done
|
|
}
|
|
|
|
(sleep 1; exit 1) &
|
|
pressure_pid=$!
|
|
AI_GATEWAY_ACCEPTANCE_PROCESS_WATCH_SECONDS=1
|
|
active_load_pid=
|
|
if run_with_pressure_monitor "$pressure_pid" long_running_load; then
|
|
echo 'load continued after the pressure sampler failed' >&2
|
|
exit 1
|
|
fi
|
|
wait "$pressure_pid" >/dev/null 2>&1 || true
|
|
[[ -z $active_load_pid ]]
|
|
|
|
deploy_call_line=$(grep -n '^deploy_protocol_emulator$' "$script" | cut -d: -f1)
|
|
activate_call_line=$(grep -n '^create_and_activate_run$' "$script" | cut -d: -f1)
|
|
[[ -n $deploy_call_line && -n $activate_call_line && $deploy_call_line -lt $activate_call_line ]]
|
|
|
|
echo 'production_acceptance_script_tests=PASS'
|