PHP-FPM WARNING: [pool www] server reached max_children setting (50), consider raising it

well tuning is up to you

keep raising it until you don't get the error or to the point you can live with the memory usage and php response times (which get slower as you get further with pm.max_children > number of cpu threads/cores available).

 

try disabling and see

 

ooh nice tweak there.. yes the 3 lines should go into php include file you listed

Code:

fastcgi_pass_request_body off;
client_body_in_file_only clean;
fastcgi_param  REQUEST_BODY_FILE  $request_body_file; 

let me know how that fairs for you

• http://nginx.org/en/docs/http/ngx_http_fastcgi_module.html#fastcgi_pass_request_body
• http://nginx.org/en/docs/http/ngx_http_core_module.html#client_body_in_file_only
• http://nginx.2469901.n2.nabble.com/Uploads-with-nginx-1-0-12-tp7330514p7330759.html

 

yeah that tweak wouldn't help fully with that error unless it's related to slow or time consuming uploads via php-fpm for your usage case. Tweak and tune = trial and error testing

let's see how it fairs for @RoldanLT first heh.

 

I'd be wary of raising max children too high - but it's all relative to how many cpu cores and cpu threads you have to work with. I mentioned that in 3rd post above https://community.centminmod.com/th...setting-50-consider-raising-it.791/#post-3607.

curious @RoldanLT which pages you get hits mostly from ? You can use ngxtop to find out like Brent did at https://community.centminmod.com/threads/ngxtop-real-time-metrics-for-nginx.285/

Code:

ngxtop -l /home/nginx/domains/yourdomain.com/log/access.log --no-follow

 

you're concerned with .php files or php files served via friendly urls which should by default still get logged to access log as you can see from Brent's stats

if you have specifically and completely disabled access_logs yourself, you can just re-enable them for 48hrs to see where that php load is coming from

 

so did you end up trying that tweak for the below 3 additional lines ?

Code:

fastcgi_pass_request_body off;
client_body_in_file_only clean;
fastcgi_param  REQUEST_BODY_FILE  $request_body_file;

 

curious to try it on this forum too

 

@RoldanLT just to elaborate on statement I made regarding usefulness of Siege for PHP tuning at Php-fpm configuration: encountered and error | Page 6 | XenForo Community

To simulate hogged PHP processes for forum attachment uploads etc (yeah not scientific), a simple example of hello world php script where one has an sleep 10 second delay to simulate PHP processing delays

with siege -c50 -d30 -t1M

###############################
dynamic

Code:

;pm = ondemand
pm = dynamic
pm.max_children = 8
; Default Value: min_spare_servers + (max_spare_servers - min_spare_servers) / 2
pm.start_servers = 6
pm.min_spare_servers = 4
pm.max_spare_servers = 8
pm.max_requests = 100

no simulated 10s sleep delay

Code:

pool:                 www
process manager:      dynamic
start time:           08/Aug/2014:20:52:57 -0700
start since:          75
accepted conn:        179
listen queue:         0
max listen queue:     0
listen queue len:     65535
idle processes:       5
active processes:     1
total processes:      6
max active processes: 3
max children reached: 0
slow requests:        0

Code:

Transactions:                    177 hits
Availability:                 100.00 %
Elapsed time:                  59.83 secs
Data transferred:               0.02 MB
Response time:                  0.00 secs
Transaction rate:               2.96 trans/sec
Throughput:                     0.00 MB/sec
Concurrency:                    0.00
Successful transactions:         177
Failed transactions:               0
Longest transaction:            0.01
Shortest transaction:           0.00

with simulated 10s sleep delay during siege

Code:

pool:                 www
process manager:      dynamic
start time:           08/Aug/2014:20:54:44 -0700
start since:          78
accepted conn:        69
listen queue:         19
max listen queue:     37
listen queue len:     65535
idle processes:       0
active processes:     8
total processes:      8
max active processes: 8
max children reached: 1
slow requests:        0

with simulated 10s sleep delay after siege

Code:

pool:                 www
process manager:      dynamic
start time:           08/Aug/2014:20:54:44 -0700
start since:          121
accepted conn:        88
listen queue:         0
max listen queue:     37
listen queue len:     65535
idle processes:       7
active processes:     1
total processes:      8
max active processes: 8
max children reached: 1
slow requests:        0

Code:

Transactions:                     39 hits
Availability:                  68.42 %
Elapsed time:                  59.32 secs
Data transferred:               0.00 MB
Response time:                 17.95 secs
Transaction rate:               0.66 trans/sec
Throughput:                     0.00 MB/sec
Concurrency:                   11.80
Successful transactions:          39
Failed transactions:              18
Longest transaction:           30.00
Shortest transaction:           0.00

Pay attention to the above max children reached, listen queue and max listen queue values during siege with simulated 10s sleep.

Siege has it's place in load and stress testing but it isn't 100% reflective of real world for PHP performance especially for forum type PHP loads which are non-cached PHP based and are not straight forward (php based attachment uploads).

 

@RoldanLT linked to Image upload (resize) not working on nginx/php-fpm | XenForo Community

Code:

With this nginx will handle the uploads, then pass the completed upload to PHP-FPM. Works great here 

I want to see via testing if that is really the case

Floren have you tried something like this ?

okay google-fu at work and officially it's called Accelerated upload support

Permissions for using "Accelerated upload support"?