my nginx stub status page says:
Active connections: 284
server accepts handled requests
28762321 28762321 183603127
Reading: 1 Writing: 6 Waiting: 277
is that many in a “waiting” state okay? or is that something I should
be worried about?
On Sun, Sep 28, 2008 at 11:05:57PM -0700, mike wrote:
my nginx stub status page says:
Active connections: 284
server accepts handled requests
28762321 28762321 183603127
Reading: 1 Writing: 6 Waiting: 277is that many in a “waiting” state okay? or is that something I should
be worried about?
Yes, this is OK. The waiting connections are keep-alive.
Here is status of one of my sites:
Active connections: 45579
server accepts handled requests
268846178 268846178 638258372
Reading: 620 Writing: 576 Waiting: 44383
What’s your HW for 50K active connection ?
Would you please paste your system configure or optimization?
There is no such system tuning sample of nginx for heavy traffic
server.
2008/9/29 Igor S. [email protected]:
On Mon, Sep 29, 2008 at 11:48:37PM +0800, Delta Y. wrote:
What’s your HW for 50K active connection ?
These are unactive keep-alive connections.
There are 5,000 req/s, mostly small static files, all are cached by VM.
Would you please paste your system configure or optimization?
There is no such system tuning sample of nginx for heavy traffic server.
FreeBSD 7.1/amd64, dual-core CPU, 4G.
/boot/loader.conf:
vm.kmem_size=1844M
decrease buff_map and pipe_map to increase KVM
64M KVA
kern.maxbcache=64M
4M KVA
kern.ipc.maxpipekva=4M
/etc/sysctl.conf:
4G, 2G KVA, 1844M KVM
797M KVA/KVM
kern.ipc.nmbjumbop=192000
504M KVA/KVM
kern.ipc.nmbclusters=229376
334M KVA/KVM
kern.ipc.maxsockets=204800
8M KVA/KVM
net.inet.tcp.maxtcptw=163840
24M KVA/KVM
kern.maxfiles=204800
kern.ipc.somaxconn=4096
On Tue, Sep 30, 2008 at 11:48:20AM +0800, Delta Y. wrote:
Great thank you!
I have put it on the nginx english wiki, the example page .
So others may take it as a reference.
However, increasing vm.kmem_size blindly will lead to unbootable
machine.
1844M is probably upper limit for FreeBSD/amd64, when other kernel maps
are decreased. 1400M is probably safe enough (by default kvm is 400M and
this is too little).
Great thank you!
I have put it on the nginx english wiki, the example page .
So others may take it as a reference.
We still need more examples on other OS and HW.
Anyone who want to report your story with NGINX can reply this thread
in the form of
OS&HW:
what operation system nginx runs on?
what hardware you use ? CPU, memory, NIC.
Purpose:
nginx works as a web server,reverse proxy or both.
Load description:
how many connections?
how many request per second?
what CPU&MEM resources nginx use?
System config:
your system config or how you tune it?
Nginx config:
your nginx config? please remember to remove the sensitive data.
2008/9/30 Igor S. [email protected]: