server_config: CONNECT 出站超时 60s→10s,并补充排障口径
实测(2026-09):目标为黑洞 IP 时,nginx 出站超时 60s > lite 客户端 15s, 表现为客户端 'read operation timed out' 而非 502,形似代理卡死,实为超时错配。 改为 10s 后黑洞目标 10.2s 返回、正常目标 0.4-0.7s 不受影响(五节点已生效)。 README 新增 5.6 节:三个超时的口径 + 'CONNECT 502' 的排查顺序 (先判断目标是否本身不可达 → 再看 nginx 错误类型 → 最后才怀疑 CONNECT 模块), 并建议客户端日志记录原始域名(案例:Dropbox 旧 IP 失效导致误判)。
This commit is contained in:
@@ -58,7 +58,9 @@ http {
|
||||
proxy_connect;
|
||||
proxy_connect_allow all;
|
||||
proxy_connect_address $rrpp_uot_connect_address;
|
||||
proxy_connect_connect_timeout 60s;
|
||||
# 出站连接超时须小于客户端超时(lite 客户端 15s):
|
||||
# 目标黑洞时快速返回 502,避免客户端"read operation timed out"干等
|
||||
proxy_connect_connect_timeout 10s;
|
||||
# 注:此模块版本 proxy_connect_data_timeout 与 proxy_connect_read_timeout
|
||||
# 写同一字段(同时配置报 duplicate),长连接超时用 read/send timeout 设置
|
||||
proxy_connect_read_timeout 10m;
|
||||
@@ -109,7 +111,9 @@ http {
|
||||
proxy_connect;
|
||||
proxy_connect_allow all;
|
||||
proxy_connect_address $rrpp_uot_connect_address;
|
||||
proxy_connect_connect_timeout 60s;
|
||||
# 出站连接超时须小于客户端超时(lite 客户端 15s):
|
||||
# 目标黑洞时快速返回 502,避免客户端"read operation timed out"干等
|
||||
proxy_connect_connect_timeout 10s;
|
||||
# 注:此模块版本 proxy_connect_data_timeout 与 proxy_connect_read_timeout
|
||||
# 写同一字段(同时配置报 duplicate),长连接超时用 read/send timeout 设置
|
||||
proxy_connect_read_timeout 10m;
|
||||
|
||||
Reference in New Issue
Block a user