# Memory consumption of ningx **URL:** \ **Category:** NGINX\ **Created:** [October 29, 2012, 6:13pm UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080 "2012-10-29T18:13:53Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![aris](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/a/258eb7/32.png) [@aris](https://www.ruby-forum.com/u/aris)\ **Post date:** [October 29, 2012, 6:13pm UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080/1 "2012-10-29T18:13:53Z")
Hi, I cross-compiled nginx for microblaze processors ([GitHub - peschuster/nginx: Fork of nginx for cross compilation changes (e.g. microblaze architecture).](http://github.com/peschuster/nginx)) and am currently doing some performance benchmarks with nginx running on a microblaze processor with a custom designed SoC on a FPGA. However, I am having problems with the memory consumption of nginx: When I perform 10,000 requests with 20 conn/s and 2 requests/conn (using httperf - 1), memory used by nginx grows to about 40 MB. When I repeat this benchmark, the used memory grows from 40 to 80 MB. The problem with this behavior is that my SoC only has 256 MB of RAM in total (the file system also runs completely from RAM using a ramdisk). Therefore nginx crashes the complete system by consuming all memory for longer/extended benchmark scenarios. Is this the intended behavior of nginx? Why isn’t it “re-using” the already allocated memory? Any hints on how I can circumvent or track down this problem? Thanks. Peter 1: httperf --timeout=5 --client=0/1 --server=192.168.2.125 --port=80 –uri=/index.html --rate=20 --send-buffer=4096 --recv-buffer=16384 –num-conns=5000 --num-calls=2 Posted at Nginx Forum: > **[Memory consumption of ningx](https://forum.nginx.org/read.php?2,232328,232328#msg-232328)** > > Hi, I cross-compiled nginx for microblaze processors (http://github.com/peschuster/nginx) and am currently doing some performance benchmarks with nginx running on a microblaze processor with a custom designed SoC on a FPGA. However, I am having... ---
**Author:** ![peschuster](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/ecae2f/32.png) [@peschuster](https://www.ruby-forum.com/u/peschuster)\ **Post date:** [October 30, 2012, 11:15am UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080/2 "2012-10-30T11:15:58Z")
On 10/29/12 21:13, peschuster wrote: > However, I am having problems with the memory consumption of nginx: > > When I perform 10,000 requests with 20 conn/s and 2 requests/conn (using > httperf - 1), memory used by nginx grows to about 40 MB. > When I repeat this benchmark, the used memory grows from 40 to 80 MB. 1. Do you use 3rd party modules? 2. This request served by nginx (e. g. static files) or proxied to some backend? 3. Memory usage depends on used features: SSL, SSI, gzip, limt rate, geo module, e. t. c. If gzip is used for static files, better to pre-compress them, and use ngx\_http\_gzip\_static\_module Also yo save memory use 1 worker and set reasonable small limit on connections: worker\_processes 1; events { worker\_connections 512; } – Anton Y. ---
**Author:** ![peschuster](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/ecae2f/32.png) [@peschuster](https://www.ruby-forum.com/u/peschuster)\ **Post date:** [October 30, 2012, 12:14pm UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080/3 "2012-10-30T12:14:31Z")
I don’t use any third-party modules, neigther SSL or gzip. nginx was compiled using the following parameters: > –with-debug > –without-http\_rewrite\_module > –without-http\_gzip\_module nginx only serves one static file. Changing the size of the file (10 KB to 200 B) has no effect on the memory comsumption of nginx. Here is my nginx.conf: > http { > include mime.types; > default\_type application/octet-stream; > > ``` > log_format main '$remote_addr - $remote_user [$time_local] > > ``` “$request” ’ > ``` > server { > listen 80; > server_name localhost; > > location / { > root html; > index index.html index.htm; > } > } > > ``` > > } Posted at Nginx Forum: [http://forum.nginx.org/read.php?2,232328,232368#msg-232368](http://forum.nginx.org/read.php?2,232328,232368#msg-232368) ---
**Author:** ![peschuster](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/ecae2f/32.png) [@peschuster](https://www.ruby-forum.com/u/peschuster)\ **Post date:** [October 30, 2012, 12:42pm UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080/4 "2012-10-30T12:42:04Z")
On Monday 29 October 2012 21:13:01 peschuster wrote: > httperf - 1), memory used by nginx grows to about 40 MB. > When I repeat this benchmark, the used memory grows from 40 to 80 MB. > > The problem with this behavior is that my SoC only has 256 MB of RAM in > total (the file system also runs completely from RAM using a ramdisk). > Therefore nginx crashes the complete system by consuming all memory for > longer/extended benchmark scenarios. > > Is this the intended behavior of nginx? Why isn’t it “re-using” the already > allocated memory? Nginx releases allocated memory after it completes each request. > Any hints on how I can circumvent or track down this problem? It most likely that your system memory allocator do not return freed memory to the OS. wbr, Valentin V. Bartenev – > **[F5 NGINX: Built for cloud-native apps. Ready for AI traffic.](https://www.f5.com/products/nginx)** > > Deliver modern applications with NGINX. Explore high-performance load balancing, API gateways, and web server solutions trusted by top global enterprises. > **[nginx: donation](http://nginx.org/en/donation.html)** ---
**Author:** ![peschuster](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/ecae2f/32.png) [@peschuster](https://www.ruby-forum.com/u/peschuster)\ **Post date:** [October 31, 2012, 3:43pm UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080/5 "2012-10-31T15:43:19Z")
On Wednesday 31 October 2012 15:28:58 peschuster wrote: > VBart Wrote: > > > It most likely that your system memory allocator do not return freed > > memory to the OS. > > How can I check this? I suspect this should be part of the OS (Linux)? Usually it’s a part of glibc in Linux. > Could you give me any keyword to read more about this? Try man mallopt. wbr, Valentin V. Bartenev – > **[F5 NGINX: Built for cloud-native apps. Ready for AI traffic.](https://www.f5.com/products/nginx)** > > Deliver modern applications with NGINX. Explore high-performance load balancing, API gateways, and web server solutions trusted by top global enterprises. > **[nginx: donation](http://nginx.org/en/donation.html)** ---
**Author:** ![peschuster](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/ecae2f/32.png) [@peschuster](https://www.ruby-forum.com/u/peschuster)\ **Post date:** [October 31, 2012, 12:29pm UTC](https://www.ruby-forum.com/t/memory-consumption-of-ningx/224080/6 "2012-10-31T12:29:16Z")
VBart Wrote: > It most likely that your system memory allocator do not return freed > memory to the OS. How can I check this? I suspect this should be part of the OS (Linux)? Could you give me any keyword to read more about this? I looked at /proc/meminfo after the first and second batch of requests. The newly allocated memory is categorized as “Active(anon)”: > Active: 36976 kB > Inactive: 59280 kB > Active(anon): 34376 kB > .. > AnonPages: 34400 kB vs. > Active: 71092 kB > Inactive: 60088 kB > Active(anon): 68428 kB > .. > AnonPages: 68452 kB Thanks. Peter Posted at Nginx Forum: [http://forum.nginx.org/read.php?2,232328,232426#msg-232426](http://forum.nginx.org/read.php?2,232328,232426#msg-232426)