I know I am adding too much information in the question but please bear with me.
We have our site running over domain say for e.g. - domain1.com having ip say for e.g. - 192.168.1.25. Now, we got another server running over IP say for e.g. 192.168.1.26 having plesk panel 12.5.30. Please note, we have not pointed our domain - domain1.com to 192.168.1.26 as we would like to keep it private until we setup everything and ready to start working as production. We have setup our domain in plesk panel with domain1.com and created staging.domain1.com as subdomain.
We modified hosts files on our local machine to point staging.domain1.com to 192.168.1.26 so we are able to access our website from new server while main domain - domain1.com is still running over 192.168.1.25 and we are able to access as domain1.com
Basically, we are setting up staging environment over 192.168.1.26 IP. We do have CentOS 6 running over new staging domain and we enabled firewall rules to run website on certain IPs in plesk panel.
We want to configure varnish over new server and we installed it. [We have already upvoted for varnish request in plesk forum.] On plesk, we do have apache running over port 7080 & nginx running over port 80.
> netstat -ntlp | grep -w 80
tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN 29795/nginx
tcp 0 0 :::80 :::* LISTEN 29795/nginx
> netstat -ntlp | grep -w 7080
tcp 0 0 0.0.0.0:7080 0.0.0.0:* LISTEN 10161/httpd
We setup A record as 192.168.1.26 - staging.domain1.com to our DNS and we are getting correct IP on staging domain. We set /etc/varnish/default.vcl file as below:
backend default {
.host = "192.168.1.26";
.port = "80";
}
backend admin {
.host = "192.168.1.26";
.port = "80";
.first_byte_timeout = 18000s;
.between_bytes_timeout = 18000s;
}
sub vcl_recv {
if (req.restarts == 0) {
if (req.http.x-forwarded-for) {
set req.http.X-Forwarded-For = req.http.X-Forwarded-For + ", " + cl$
} else {
set req.http.X-Forwarded-For = client.ip;
}
}
if (req.request != "GET" &&
req.request != "HEAD" &&
req.request != "PUT" &&
req.request != "POST" &&
req.request != "TRACE" &&
req.request != "OPTIONS" &&
req.request != "DELETE" &&
req.request != "PURGE") {
/* Non-RFC2616 or CONNECT which is weird. */
return (pipe);
}
if (req.http.cookie ~ "adminhtml=") {
set req.backend = admin;
}
}
Here is the /etc/sysconfig/varnish file:
VARNISH_LISTEN_PORT=6081
VARNISH_ADMIN_LISTEN_PORT=6082
VARNISH_VCL_CONF=/etc/varnish/default.vcl
We started varnish service - /etc/init.d/varnish start and checked port 6081:
> netstat -ntlp | grep -w 6081
tcp 0 0 0.0.0.0:6081 0.0.0.0:* LISTEN 23745/varnishd
tcp 0 0 :::6081 :::* LISTEN 23745/varnishd
But when we check our site by accessing staging.domain1.com:6081, we are getting response but URLs showing in browser are pointing at domain1.com. So we changed .host in default & admin to domain name - staging.domain1.com such as below and then restarted varnish service but it is failing to do so.
backend default {
.host = "staging.domain1.com";
.port = "80";
}
backend admin {
.host = "staging.domain1.com";
.port = "80";
.first_byte_timeout = 18000s;
.between_bytes_timeout = 18000s;
}
I am not getting the reason why it is not restarting after adding real (sub)domain names. Another thing: Site is serving from port 80 i.e. nginx. Ideally, varnish should be running over port 80 and we tried to do so but somehow we failed. We managed to run nginx over port 8080 by following guidelines at - https://www.theshell.guru/change-nginx-port-plesk-12-centos-6-6/. Basically, we added custom directory under /usr/local/psa/admin/conf/templates and copied following files over there and updated to port 8080.
- nginx.php
- nginxDomainForwarding.php
- nginxDomainForwardingIpDefault.php
- nginxDomainVhost.php
- nginxDomainVhostIpDefault.php
- nginxWebmail.php
Then we changed, varnish listen port to 80 instead of port 6081 but when we start varnish, it straightaway shows failed and our domain - staging.domain1.com fails to load so we reverted back everything as above i.e.
- varnish on port 6081[VCL backend at 80]
- apache 7080
- nginx 80
I assumed I should have following scenario:
- Varnish - port 80 [default.vcl file should greb content from 8080 port]
- Nginx - port 8080
- Apache - port 7080.
In first scenario, we are not able to confirm whether varnish is correctly configured or not [in case if we set IP instead of real host name] and in second scenario, we are not able to start varnish service though it should be ideal case.
Basically, I am not getting what is wrong over here.
