# Trying to Understand Upstream Keepalive

**URL:** <https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966>\
**Category:** NGINX\
**Created:** [May 8, 2014, 9:12am UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966 "2014-05-08T09:12:59Z")\
**Posts on this page:** 10\
**Page:** 1

<div class="post-metadata">

**Author:** ![Detlef\_R](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/58f4c7/32.png) [@Detlef\_R](https://www.ruby-forum.com/u/Detlef_R)\
**Post date:** [May 8, 2014, 9:12am UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/1 "2014-05-08T09:12:59Z")

</div>

I’m trying to better wrap my head around the keepalive functionality in  
the  
upstream module as when enabling keepalive, I’m seeing little to no  
performance benefits using the FOSS version of nginx.

My upstream block is:

upstream upstream\_test\_1 { server 1.1.1.1 max\_fails=0; keepalive 50; }

With a proxy block of:

proxy\_set\_header X-Forwarded-For $IP;  
proxy\_set\_header Host $http\_host;  
proxy\_http\_version 1.1;  
proxy\_set\_header Connection “”;  
proxy\_pass http://upstream\_test\_1;

1. How can I tell whether there are any connections currently in the  
keepalive pool for the upstream block? My origin server has keepalive  
enabled and I see that there are some connections in a keepalive state,  
however not the 50 defined and all seem to close much quicker than the  
keepalive timeout for the backend server. (I am using the Apache server  
status module to view this which is likely part of the problem)

2. Are upstream blocks shared across workers? So in this situation,  
would  
all 4 workers I have shared the same upstream keepalive pool or would  
each  
worker have it’s own block of 50?

3. How is the length of the keepalive determined? The origin server’s  
keepalive settings? Do the origin server’s keepalive settings factor in  
at  
all?

4. If no traffic comes across this upstream for an extended period of  
time,  
will the connections be closed automatically or will they stay open  
infinitely?

5. Are the connections in keepalive shared across visitors to the proxy?  
For  
example, if I have three visitors to the proxy one after the other,  
would  
the expectation be that they use the same connection via keepalive or  
would  
a new connection be opened for each of them?

6. Is there any common level of performance benefit I should be seeing  
from  
enabling keepalive compared to just performing a proxy\_pass directly to  
the  
origin server with no upstream block?

Thanks for any insight!

Posted at Nginx Forum:

> **[Trying to Understand Upstream Keepalive](https://forum.nginx.org/read.php?2,249924,249924#msg-249924)**
>
> I'm trying to better wrap my head around the keepalive functionality in the upstream module as when enabling keepalive, I'm seeing little to no performance benefits using the FOSS version of nginx. My upstream block is: upstream upstream\_test\_1 {...

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [May 9, 2014, 4:08am UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/2 "2014-05-09T04:08:26Z")

</div>

Hello!

On Thu, May 08, 2014 at 03:12:44AM -0400, abstein2 wrote:

> proxy\_set\_header X-Forwarded-For $IP;  
> status module to view this which is likely part of the problem)  
> As long as load is even enough, don’t expect to see many keepalive  
> connections on the backend - new connections will be only open if  
> there are no idle connections in the cache of a worker process.

> 1. Are upstream blocks shared across workers? So in this situation, would  
> all 4 workers I have shared the same upstream keepalive pool or would each  
> worker have it’s own block of 50?

It’s per worker, see [Module ngx\_http\_upstream\_module](http://nginx.org/r/keepalive).

> 1. How is the length of the keepalive determined? The origin server’s  
> keepalive settings? Do the origin server’s keepalive settings factor in at  
> all?

Connections are kept in the cache till the origin server closes them.

> 1. If no traffic comes across this upstream for an extended period of time,  
> will the connections be closed automatically or will they stay open  
> infinitely?

See above.

> 1. Are the connections in keepalive shared across visitors to the proxy? For  
> example, if I have three visitors to the proxy one after the other, would  
> the expectation be that they use the same connection via keepalive or would  
> a new connection be opened for each of them?

Connections in the cache are shared for all uses of the upstream.  
As long as a connection is idle (and hence in the cache), it can  
by used for any request by any visitor.

> 1. Is there any common level of performance benefit I should be seeing from  
> enabling keepalive compared to just performing a proxy\_pass directly to the  
> origin server with no upstream block?

No.

There are two basic cases when keeping connections alive is  
really beneficial:

- Fast backends, which produce responses is a very short time,  
comparable to a TCP handshake.

- Distant backends, when a TCP handshake takes a long time,  
comparable to a backend response time.

There are also some bonus side effects (reducing number of sockets  
in TIME-WAIT state, less work for OS to establish new connections,  
less packets on a network), but these are unlikely to result in  
measurable performance benefits in a typical setup.

–  
Maxim D.  
[http://nginx.org/](http://nginx.org/)

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [October 29, 2014, 1:01am UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/3 "2014-10-29T01:01:49Z")

</div>

Maxim,

[http://nginx.org/en/docs/http/ngx\_http\_upstream\_module.html#keepalive](http://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive)

I would like to know what is the keepalive timeout for this connection  
pool?  
Is it static?

Also i want to understand - if there is a marriage between number of  
connections nginx gets vs how many it opens to upstream?

Thanks!

Posted at Nginx Forum:

> **[Re: Trying to Understand Upstream Keepalive](https://forum.nginx.org/read.php?2,249924,254393#msg-254393)**
>
> I'm trying to better wrap my head around the keepalive functionality in the upstream module as when enabling keepalive, I'm seeing little to no performance benefits using the FOSS version of nginx. My upstream block is: upstream upstream\_test\_1 {...

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [October 29, 2014, 6:16pm UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/4 "2014-10-29T18:16:31Z")

</div>

## Maxim D. Wrote:

> connection pool?
> 
> > Is it static?
> 
> As of now, there is no timeout on nginx side. Connections are  
> closed either by backends or if there isn’t enough room in  
> the cache.

So how long after a connection to upstream goes from ACTIVE to idle in  
the  
connection pool does it get closed?  
There is not really much documentation on this upstream keepalive  
component.

> > Also i want to understand - if there is a marriage between number of  
> > connections nginx gets vs how many it opens to upstream?
> 
> This depends on how long it takes to process a request (as well as  
> various other factors). As long as backends are fast enough, one  
> connection to upstream may be enough to handle tens or hundreds of  
> client connections.

Ok.  
What is the difference between ‘max\_conns’ vs  
[http://nginx.org/en/docs/http/ngx\_http\_limit\_conn\_module.html#limit\_conn](http://nginx.org/en/docs/http/ngx_http_limit_conn_module.html#limit_conn)  
to  
an upstream service ?

> –  
> Maxim D.  
> [http://nginx.org/](http://nginx.org/)
> 
> * * *
> 
> nginx mailing list  
> removed\_email\_address@domain.invalid  
> [nginx Info Page](http://mailman.nginx.org/mailman/listinfo/nginx)

Posted at Nginx Forum:

> **[Re: Trying to Understand Upstream Keepalive](https://forum.nginx.org/read.php?2,249924,254418#msg-254418)**
>
> I'm trying to better wrap my head around the keepalive functionality in the upstream module as when enabling keepalive, I'm seeing little to no performance benefits using the FOSS version of nginx. My upstream block is: upstream upstream\_test\_1 {...

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [October 29, 2014, 5:30pm UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/5 "2014-10-29T17:30:34Z")

</div>

Hello!

On Tue, Oct 28, 2014 at 08:01:33PM -0400, newnovice wrote:

> Maxim,
> 
> [Module ngx\_http\_upstream\_module](http://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive)
> 
> I would like to know what is the keepalive timeout for this connection pool?  
> Is it static?

As of now, there is no timeout on nginx side. Connections are  
closed either by backends or if there isn’t enough room in  
the cache.

> Also i want to understand - if there is a marriage between number of  
> connections nginx gets vs how many it opens to upstream?

This depends on how long it takes to process a request (as well as  
various other factors). As long as backends are fast enough, one  
connection to upstream may be enough to handle tens or hundreds of  
client connections.

–  
Maxim D.  
[http://nginx.org/](http://nginx.org/)

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [October 29, 2014, 7:20pm UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/6 "2014-10-29T19:20:58Z")

</div>

“isn’t enough room in the cache.”

how big is the upstream keepalive connection-pool cache size?

Posted at Nginx Forum:

> **[Re: Trying to Understand Upstream Keepalive](https://forum.nginx.org/read.php?2,249924,254426#msg-254426)**
>
> I'm trying to better wrap my head around the keepalive functionality in the upstream module as when enabling keepalive, I'm seeing little to no performance benefits using the FOSS version of nginx. My upstream block is: upstream upstream\_test\_1 {...

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [October 29, 2014, 9:14pm UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/7 "2014-10-29T21:14:58Z")

</div>

Hello!

On Wed, Oct 29, 2014 at 02:20:28PM -0400, newnovice wrote:

> “isn’t enough room in the cache.”
> 
> how big is the upstream keepalive connection-pool cache size?

[http://nginx.org/en/docs/http/ngx\_http\_upstream\_module.html#keepalive](http://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive)

–  
Maxim D.  
[http://nginx.org/](http://nginx.org/)

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [October 29, 2014, 6:53pm UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/8 "2014-10-29T18:53:30Z")

</div>

Hello!

On Wed, Oct 29, 2014 at 01:15:56PM -0400, newnovice wrote:

> > >
> 
> There is not really much documentation on this upstream keepalive component.  
> That’s unspecified, see above.

> [Module ngx\_http\_limit\_conn\_module](http://nginx.org/en/docs/http/ngx_http_limit_conn_module.html#limit_conn) to  
> an upstream service ?

The “max\_conns” parameter (only available in nginx-plus) limits  
the number of active connections to an upstream server, while  
limit\_conn limits the number of active connections to a particular  
location. This difference may be significant, for example, in the  
following cases:

- there are many upstream servers in a single upstream{} block;

- some responses are returned from cache;

- responses are large enough and clients are slow, so responses  
are buffered by nginx for a long time.

–  
Maxim D.  
[http://nginx.org/](http://nginx.org/)

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [June 14, 2016, 10:09am UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/9 "2016-06-14T10:09:22Z")

</div>

As stated in  
[http://nginx.org/en/docs/http/ngx\_http\_upstream\_module.html#keepalive](http://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive)  
that  
“It should be particularly noted that the keepalive directive does not  
limit  
the total number of connections to upstream servers that an nginx worker  
process can open. The connections parameter should be set to a number  
small  
enough to let upstream servers process new incoming connections as  
well.” I  
want to understand if a new client comes, why can’t they use existing  
keep-alive connections? Do they need to create a new connection with  
upstream?

Posted at Nginx Forum:

> **[Re: Trying to Understand Upstream Keepalive](https://forum.nginx.org/read.php?2,249924,267553#msg-267553)**
>
> I'm trying to better wrap my head around the keepalive functionality in the upstream module as when enabling keepalive, I'm seeing little to no performance benefits using the FOSS version of nginx. My upstream block is: upstream upstream\_test\_1 {...

---

<div class="post-metadata">

**Author:** ![abstein2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/a3d4f5/32.png) [@abstein2](https://www.ruby-forum.com/u/abstein2)\
**Post date:** [June 14, 2016, 2:13pm UTC](https://www.ruby-forum.com/t/trying-to-understand-upstream-keepalive/240966/10 "2016-06-14T14:13:33Z")

</div>

On Tuesday 14 June 2016 04:09:06 aanchalj wrote:

> As stated in  
> [Module ngx\_http\_upstream\_module](http://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive) that  
> “It should be particularly noted that the keepalive directive does not limit  
> the total number of connections to upstream servers that an nginx worker  
> process can open. The connections parameter should be set to a number small  
> enough to let upstream servers process new incoming connections as well.” I  
> want to understand if a new client comes, why can’t they use existing  
> keep-alive connections? Do they need to create a new connection with  
> upstream?

[…]

It’s about the case when all of the existing keep-alive connections are  
already  
in use and processing other client requests.

wbr, Valentin V. Bartenev