Lots of questions:

What are the upstream requests?
Are you logging hits and misses for the cache - what's the hit ratio?
What size are the objects that you are serving?
How many files are there in your cache?
What OS and what hardware are you using? If it's Linux can you show the results of the following:
cat /proc/cpuinfo | tail -30
cat /proc/meminfo

Sent from my iPhone

> On Sep 20, 2017, at 7:34 PM, Nick Urbanik <nick.urbanik@optusnet.com.au> wrote:
>
> Dear Folks,
>
>> On 19/09/17 11:16 +1000, Nick Urbanik wrote:
>>> On 19/09/17 10:46 +1000, Nick Urbanik wrote:
>>> We have this message repeatedly, despite increasing keys_zone size by
>>> a factor of three to:
>>>
>>> proxy_cache_path /srv/mycache levels=1:2 keys_zone=myzone:150m inactive=15d;
>
> Is it possible that we are way underspecifying the shared memory in keys_zone?
> --
> Nick Urbanik http://nicku.org 808-71011 nick.urbanik@optusnet.com.au
> GPG: 7FFA CDC7 5A77 0558 DC7A 790A 16DF EC5B BB9D 2C24 ID: BB9D2C24
> I disclaim, therefore I am.
> _______________________________________________
> nginx mailing list
> nginx@nginx.org
> http://mailman.nginx.org/mailman/listinfo/nginx
_______________________________________________
nginx mailing list
nginx@nginx.org
http://mailman.nginx.org/mailman/listinfo/nginx