# How to wait for upstream server to start

**URL:** <https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020>\
**Category:** NGINX\
**Created:** [February 23, 2010, 11:03pm UTC](https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020 "2010-02-23T23:03:44Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![kevin\_gill](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/k/f19dbf/32.png) [@kevin\_gill](https://www.ruby-forum.com/u/kevin_gill)\
**Post date:** [February 23, 2010, 11:03pm UTC](https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020/1 "2010-02-23T23:03:44Z")

</div>

I am running a low traffic site using nginx. I use one upstream  
multi-threadded server to serve the content. I use the proxy module to  
forward the requests.

Occasionally, I need to restart the upstream server. I takes 20-30  
seconds  
to startup.

I want requests to queue until the server is started, i.e. configure a  
30  
second timeout before I get the “502 Bad Gateway” response.

I am running nginx/0.7.64 on ubuntu.

Is there an nginx configuration option to make it wait until and retry  
while the server restarts? Alternatively, is there another product I  
should put in the middle between nginx and my upstream server which  
provides this functionality.

Thanks,

Kevin

---

<div class="post-metadata">

**Author:** ![kevin\_gill](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/k/f19dbf/32.png) [@kevin\_gill](https://www.ruby-forum.com/u/kevin_gill)\
**Post date:** [February 24, 2010, 6:02pm UTC](https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020/2 "2010-02-24T18:02:21Z")

</div>

I got this working using haproxy with the ‘dispatch’ option.

Thanks

---

<div class="post-metadata">

**Author:** ![kevin\_gill](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/k/f19dbf/32.png) [@kevin\_gill](https://www.ruby-forum.com/u/kevin_gill)\
**Post date:** [February 24, 2010, 6:24pm UTC](https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020/3 "2010-02-24T18:24:00Z")

</div>

could you describe how you got this working? i need something similar  
to make sure that a django app doesn’t die.

–timball

On Tue, Feb 23, 2010 at 6:59 PM, kevin gill [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)  
wrote:

> > to startup.
> 
> * * *
> 
> nginx mailing list  
> [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)  
> [mailman.nginx.org archive](http://nginx.org/mailman/listinfo/nginx)

–  
GPG key available on pgpkeys.mit.edu  
pub 1024D/511FBD54 2001-07-23 Timothy Lu Hu Ball [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)  
Key fingerprint = B579 29B0 F6C8 C7AA 3840 E053 FE02 BB97 511F BD54

---

<div class="post-metadata">

**Author:** ![kevin\_gill](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/k/f19dbf/32.png) [@kevin\_gill](https://www.ruby-forum.com/u/kevin_gill)\
**Post date:** [February 25, 2010, 4:30am UTC](https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020/4 "2010-02-25T04:30:15Z")

</div>

> could you describe how you got this working? i need something similar  
> to make sure that a django app doesn’t die.

I assume that you are modifing a working nginx/django configuration.

Mine is a zope server. For static content I go to varnish and for the  
rest  
I use proxy\_pass to go to zope. I inserted haproxy in front of zope.

nginx → varnish → haproxy → zope  
nginx ------------\> haproxy → zope

nginx is on port 80  
varnish is on port 8082  
zope is on port 8081  
haproxy is on port 8085

#------------------------------------------------------------  
nginx server configuration…

upstream varnish1 {  
server 127.0.0.1:8082;  
}  
upstream haproxy {  
server 127.0.0.1:8085;  
}

#------------------------------------------------------------  
haproxy configuration…

global  
log 127.0.0.1 local0  
log 127.0.0.1 local1 notice  
maxconn 4096  
user haproxy  
group haproxy  
daemon

```
# enable debug to run in the foreground
debug

```

defaults  
mode http  
retries 100  
option redispatch  
timeout connect 60s

```
# I have some very slow requests - generate 1000's of emails, 

```

printouts  
timeout queue 300s  
timeout client 1200s  
timeout server 1200s

```
# this is just monitoring stuff
monitor-uri /haproxy-ping
stats enable
stats uri /haproxy-status
stats refresh 5s
stats realm Haproxy statistics

```

listen zope3 0.0.0.0:8085  
dispatch 127.0.0.1:8081 maxcon 3

#------------------------------------------------------------  
Install haproxy…

$ apt-get install haproxy  
edit /etc/default/haproxy and enable it  
$ /etc/init.d/haproxy start

Regards,

Kevin

---

<div class="post-metadata">

**Author:** ![kevin\_gill](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/k/f19dbf/32.png) [@kevin\_gill](https://www.ruby-forum.com/u/kevin_gill)\
**Post date:** [February 24, 2010, 10:35pm UTC](https://www.ruby-forum.com/t/how-to-wait-for-upstream-server-to-start/184020/5 "2010-02-24T22:35:39Z")

</div>

Hi Kevin,

kevin gill a écrit :

> I want requests to queue until the server is started, i.e. configure a 30  
> second timeout before I get the “502 Bad Gateway” response.

I’m pretty sure that you can do this with proxy\_connect\_timeout:

> **[GitHub - nginxinc/nginx-wiki: ARCHIVED -- Source for the now archived NGINX Wiki...](https://github.com/nginxinc/nginx-wiki)**
>
> ARCHIVED -- Source for the now archived NGINX Wiki section of https://www.nginx.com

Hope this helps,  
GFK’s