Yichun Zhang

Some suggestions: 1. Check out your nginx's error.log file to see if there's any errors or if Nginx is flushing too much data to it. Flushing error logs is expensive....

How many CPU cores have you assigned to your VMWare instance? How many cores are detected in your guest OS? Are you sure your worker_cpu_affinity directive is configured in the...

And yeah, you're expected to observe significant performance drop in a virtual machine especially when you're not using the hardware emulation technology :)

It seems that you're not following all my previous suggestions. Please do that. Especially: - Use the flamegraph tool to analyze how the CPU time is distributed in your Nginx...

@kilgor It's a common mistake. In a single location, only one nginx module is allowed to run as the content handler to generate the response. Here you have two modules...

@kilgor `add_header` does not work as a "content handler", that is, it does not generate a response from scratch, rather, it runs in the "output header filter" phase which just...

@kilgor No, you don't really need a `redis2_slient_query` directive here. Otherwise you'll quickly end up with a `xxx_silent_xxx` variant for literally _every_ existing directive of all those nginx content handler...

@derekzhan Looks like a low-level network failure in your box.

I'm terribly sorry for the delay! I don't know why I missed the email for this ticket :( ngx_redis2 only supports nginx variables pointing to upstream names that are already...

@signals2 There has been a test case for this usage in ngx_redis2's test suite that keeps passing for long: https://github.com/openresty/redis2-nginx-module/blob/master/t/sanity.t#L294 One common mistake is that you forget to define the...