Re: Port Exhaustion - SQL

2019-05-16 Thread Brandon Mallory
That is what I was thinking, I am having an issue with the listen directive, what should I use since the local port is "random" also for split_clients "$remote_addr$remote_port" $split_ip I cant use remote address since its a local address ?? same with port ? This is what I have been trying

RE: Port Exhaustion - SQL

2019-05-16 Thread Reinis Rozitis
> Yes all of those changes you have mentioned have been made. Well imo there is nothing else besides to even more decrease the FIN timeout (in a LAN that shouldn't be an issue (no slow clients)) so the lingering sockets are closed faster. Also instead of adding the network adapter(s) on the

Port Exhaustion - SQL

2019-05-16 Thread CCS
I have recently moved our sql server off of our webserver and now we are experiencing Port Exhaustion. We have made all the changed we could in the kernel to help with this but still hitting limits. I have added a 2nd virtual network adapter, I am now trying to "load balance" the lan connection

Re: Port Exhaustion - SQL

2019-05-16 Thread Brandon Mallory
Yes all of those changes you have mentioned have been made. Thanks Brandon Best Regards, Brandon Mallory Network & Systems Engineer MEDENT EMR/EHR 15 Hulbert Street Auburn, NY 13021 Phone: [ callto:(315)-255-0900 | (315)-255-0900 ] Fax: [ callto:(315)-255-3539 | (315)-255-3539 ] Web:

RE: Port Exhaustion - SQL

2019-05-16 Thread Reinis Rozitis
> I love nginx and use it for other applications but maybe its the wrong > product for this senerio Does nginx connect to mysql (like you use some kind of embedded module (perl/lua etc)?) or do you proxy some backend app? If not then it has no relation to this issue. > We do not have an issue

Re: Port Exhaustion - SQL

2019-05-16 Thread Maxim Konovalov
On 16/05/2019 16:51, Maxim Konovalov wrote: > Hi, > > On 16/05/2019 16:46, Brandon Mallory wrote: >> This is a very busy server and tried to push our programming >> department to move to persistent connections, they feel that it >> could be a security issue if dealing with sensitive information

RE: Port Exhaustion - SQL

2019-05-16 Thread Reinis Rozitis
> We have made all the changed we could in the kernel to help with this but > still hitting limits. What changes have you made? Usually the port limit is reached because of time wait sockets. If not done already try with: net.ipv4.ip_local_port_range = 1028 65535 net.ipv4.tcp_tw_reuse = 1

Re: Port Exhaustion - SQL

2019-05-16 Thread Brandon Mallory
This is a very busy server and tried to push our programming department to move to persistent connections, they feel that it could be a security issue if dealing with sensitive information since that connection could be hijacked. We do not have an issue on the mysql server side with Port

Re: Port Exhaustion - SQL

2019-05-16 Thread Maxim Konovalov
Hi, On 16/05/2019 16:46, Brandon Mallory wrote: > This is a very busy server and tried to push our programming > department to move to persistent connections, they feel that it > could be a security issue if dealing with sensitive information > since that connection could be hijacked. We do not

RE: Port Exhaustion - SQL

2019-05-16 Thread Reinis Rozitis
Ohh I missed the whole idea that nginx is used as tcp balancer for mysql. But imo it is still more simple (unless you can't do anything with the DB server) to balance the remote server rather than split and bind local clients: upstream backend { least_conn; server ip1:3306; server ip2:3306;

Re: Port Exhaustion - SQL

2019-05-16 Thread Brandon Mallory
The programmers currently use a file to specify the IP of the remote SQL server. As a work around for the time being I have added a 2nd interface to the front end server and the remote SQL server. These are the errors we get when we hit the 65k limit. mysqli_connect(): Can't connect to MySQL

[njs] Fixed uninitialized-memory-access in Object.defineProperties().

2019-05-16 Thread Dmitry Volyntsev
details: https://hg.nginx.org/njs/rev/e0fdef4eb478 branches: changeset: 965:e0fdef4eb478 user: Dmitry Volyntsev date: Thu May 16 15:20:31 2019 +0300 description: Fixed uninitialized-memory-access in Object.defineProperties(). This closes #158 issue on Github. diffstat:

Re: redirect after body capture

2019-05-16 Thread Maxim Dounin
Hello! On Wed, May 15, 2019 at 05:35:54PM -0700, Dk Jack wrote: > Hi, > In my module, I am trying to forward the request to my server based on the > content of the request body. To acheive this, I've added a body capture > filter to capture the body. Here are the relevant parts of my code... >

[njs] Improved String.prototype.replace().

2019-05-16 Thread Alexander Borisov
details: https://hg.nginx.org/njs/rev/58de3054b28a branches: changeset: 967:58de3054b28a user: Alexander Borisov date: Tue May 14 13:00:44 2019 +0300 description: Improved String.prototype.replace(). If replace value is a function and the function return non string value, then

[njs] Fixed processing * quantifier for String.prototype.replace().

2019-05-16 Thread Alexander Borisov
details: https://hg.nginx.org/njs/rev/9005991a53d6 branches: changeset: 966:9005991a53d6 user: Alexander Borisov date: Wed May 08 17:05:05 2019 +0300 description: Fixed processing * quantifier for String.prototype.replace(). This closes #73 issue on GitHub. diffstat:

[nginx] SSL: removed OpenSSL 0.9.7 compatibility.

2019-05-16 Thread Sergey Kandaurov
details: https://hg.nginx.org/nginx/rev/b99cbafd51da branches: changeset: 7509:b99cbafd51da user: Sergey Kandaurov date: Mon Apr 11 15:46:36 2016 +0300 description: SSL: removed OpenSSL 0.9.7 compatibility. diffstat: src/event/ngx_event_openssl.c| 47