非常感谢您的帮助。我的设置：LB -> VARNISH -> NGINX
由于未知原因，我开始在客户端上收到随机错误 503。我检查了 varnish 和 Nginx 上的日志...
Varnish：在日志中我看到客户端请求，RespStatus 503，ResReason Backend 获取失败。
Nginx：如果我在 Nginx 日志中搜索相同的客户端请求，则看不到它。
我猜问题应该出在 Nginx 服务器部分？我查看了网络流量/CPU/内存等... 看起来没有出现过载。
我的 Nginx 配置：
user www-data;
pid /run/nginx.pid;
worker_processes auto;
worker_rlimit_nofile 65535;
pcre_jit on;
error_log /var/log/nginx/error.log error;
include /etc/nginx/modules/*.conf;
events {
worker_connections 65535;
}
http {
include /etc/nginx/sites-available/*;
include /etc/nginx/mime.types;
default_type application/octet-stream;
server_tokens off;
client_max_body_size 1m;
sendfile on;
tcp_nopush on;
tcp_nodelay on;
keepalive_timeout 65;
types_hash_max_size 2048;
ssl_protocols TLSv1.1 TLSv1.2 TLSv1.3;
ssl_dhparam /etc/ssl/nginx/dh2048.pem;
ssl_prefer_server_ciphers on;
ssl_session_cache shared:SSL:2m;
ssl_session_timeout 1h;
ssl_session_tickets off;
gzip on;
gzip_vary on;
gzip_proxied any;
gzip_comp_level 6;
gzip_buffers 16 8k;
gzip_http_version 1.1;
gzip_disable "msie6";
gzip_types text/plain text/css application/json application/javascript text/xml application/xml application/xml+rss text/javascript application/vnd.apple.mpegurl;
map $http_upgrade $connection_upgrade {
default upgrade;
'' close;
}
log_format main '$remote_addr - $remote_user [$time_local] "$request" '
'$status $body_bytes_sent "$http_referer" '
'"$http_user_agent" "$http_x_forwarded_for"';
}
漆：
backend server1 {
.host = "192.168.245.167";
.port = "80";
.max_connections = 3000;
.first_byte_timeout = 300s;
.connect_timeout = 10s;
.between_bytes_timeout = 6s;
}
sub vcl_init {
# Called when VCL is loaded, before any requests pass through it.
# Typically used to initialize VMODs.
new vdir = directors.round_robin();
vdir.add_backend(server1);
...
编辑：
还为 1 个请求添加 varnislog 输出：
* << Request >> 609730940
- Begin req 640024598 rxreq
- Timestamp Start: 1661355668.683547 0.000000 0.000000
- Timestamp Req: 1661355668.683547 0.000000 0.000000
- VCL_use boot
- ReqStart 192.168.245.165 40772 a0
- ReqMethod GET
- ReqURL someurl
- ReqProtocol HTTP/1.1
- ReqHeader user-agent: agent
- ReqHeader accept-encoding: gzip
- ReqHeader host: example.com
- ReqHeader x-real-ip: 172.31.68.46
- ReqHeader x-ssl: 1
- ReqHeader x-forwarded-for: 172.31.68.46
- ReqUnset x-forwarded-for: 172.31.68.46
- ReqHeader X-Forwarded-For: 172.31.68.46, 192.168.245.165
- VCL_call RECV
- ReqUnset X-Forwarded-For: 172.31.68.46, 192.168.245.165
- ReqHeader X-Forwarded-For: 172.31.68.46
- ReqUnset host: example.com
- ReqHeader Host: example.com
- ReqURL someurl
- VCL_return pass
- VCL_call HASH
- VCL_return lookup
- VCL_call PASS
- VCL_return fetch
- Link bereq 609730941 pass
- Timestamp Fetch: 1661355668.683759 0.000212 0.000212
- RespProtocol HTTP/1.1
- RespStatus 503
- RespReason Backend fetch failed
- RespHeader Date: Wed, 24 Aug 2022 15:41:08 GMT
- RespHeader Server: Varnish
- RespHeader Content-Type: text/html; charset=utf-8
- RespHeader Retry-After: 5
- RespHeader X-Varnish: 609730940
- RespHeader Age: 0
- RespHeader Via: 1.1 varnish (Varnish/6.5)
- VCL_call DELIVER
- RespHeader X-Cache: MISS
- RespHeader X-Cache-Hits: 0
- VCL_return deliver
- Timestamp Process: 1661355668.683778 0.000231 0.000019
- Filters
- RespHeader Content-Length: 286
- RespHeader Connection: keep-alive
- Timestamp Resp: 1661355668.683838 0.000291 0.000060
- ReqAcct 429 0 429 286 286 572
- End
** << BeReq >> 609730941
-- Begin bereq 609730940 pass
-- VCL_use boot
-- Timestamp Start: 1661355668.683670 0.000000 0.000000
-- BereqMethod GET
-- BereqURL someurl
-- BereqProtocol HTTP/1.1
-- BereqHeader user-agent: agent
-- BereqHeader accept-encoding: gzip
-- BereqHeader x-real-ip: 172.31.68.46
-- BereqHeader x-ssl: 1
-- BereqHeader X-Forwarded-For: 172.31.68.46
-- BereqHeader Host: example.com
-- BereqHeader X-Varnish: 609730941
-- VCL_call BACKEND_FETCH
-- VCL_return fetch
-- FetchError backend server1: busy
-- Timestamp Beresp: 1661355668.683706 0.000036 0.000036
-- Timestamp Error: 1661355668.683709 0.000038 0.000002
-- BerespProtocol HTTP/1.1
-- BerespStatus 503
-- BerespReason Backend fetch failed
-- BerespHeader Date: Wed, 24 Aug 2022 15:41:08 GMT
-- BerespHeader Server: Varnish
-- VCL_call BACKEND_ERROR
-- BerespHeader Content-Type: text/html; charset=utf-8
-- BerespHeader Retry-After: 5
-- VCL_return deliver
-- Storage malloc Transient
-- Length 286
-- BereqAcct 0 0 0 0 0 0
-- End
关于我应该寻找什么，有什么想法/提示吗？
答案1
该-- FetchError backend server1: busy消息表明 Varnish 无法连接到 Nginx，因为 Nginx忙碌的。这意味着它已达到其最大连接数限制。
.max_connections如果您认为服务器可以处理更多同时连接，请在 Nginx 中增加此限制，或者通过降低后端定义中的属性来限制 Varnish 可以向 Nginx 打开的后端连接数量。
但是限制 Varnish 中的最大连接数并不能解决这个问题。
弄清楚为什么 Varnish 需要打开这么多连接是很重要的。
- 命中率是不是太低了？
- 对不可缓存的资源的请求是否太多？
- 你的 VCL 中是否存在缺陷，导致后端服务器过载
我建议你读一下https://www.varnish-software.com/developers/tutorials/troubleshooting-varnish/#varnish-is-not-caching了解如何发现不可缓存的内容。根据日志，您可以进行评估，并可能在 VCL 中采取纠正措施。
我还建议您阅读以下教程：https://www.varnish-software.com/developers/tutorials/varnish-builtin-vcl/。它解释了 Varnish 默认缓存的内容以及哪些内容被视为不可缓存。这将帮助您了解需要在多大程度上进行潜在的 VCL 改进。
答案2
谢谢你的回答。问题在于连接限制，因为我有很多文件没有缓存。也谢谢你的链接……我们会看看如何改进我们的配置。
BR，罗伯特