Re: [rt-users] spawn-fcgi crashing, is uWSGI an option?
Hi Aaron, I'm running Nginx with uWsgi in my lab environment and it seems to we working well. Here are the snippets of config I'm using. server { root /usr/share/nginx/www; index index.html index.htm; # Make site accessible from http://localhost/ server_name localhost; location / { uwsgi_pass unix:/var/tmp/uwsgi.sock; include uwsgi_params; uwsgi_modifier1 5; } } [uwsgi] plugins = psgi chmod-socket = 664 uid = www-data gid = www-data socket = /var/tmp/uwsgi.sock psgi = /opt/rt4/sbin/rt-server processes = 10 Best Regards Martin On 2015-05-31 05:43, Aaron C. de Bruyn wrote: Thanks Ken. It's working perfectly. -A On Sat, May 30, 2015 at 5:00 PM, k...@rice.edu k...@rice.edu wrote: On Sat, May 30, 2015 at 04:49:00PM -0700, Aaron C. de Bruyn wrote: I have a demo of RT set up for my company. During testing, we found the spawn-fcgi process is slow and crashes often. If more than a few users are on the site, and/or more than few messages coming in through rt-mailgate, the site loads extremely slow. We increased the number of child processes for spawn-fcgi, but it's doesn't seem adequate to handle the load. We can regularly make spawn-fcgi die by simply hitting escape while a page is loading. Other times it crashes for no apparent reason. We use uWSGI everywhere for python projects we host, and I tried to see if I could get RT working under that. I failed miserably because I don't know that much about PSGI, FCGI, etc...and their relationship to the webserver. I know Nginx can speak the uwsgi protocol to talk to uWSGI, but uWSGI seems to bomb on trying to serve RT. I tried the following config to no avail: [uwsgi] plugins = psgi socket = 127.0.0.1:9001 psgi = /opt/rt4/sbin/rt-server.fcgi fastcgi-socket = true processes = 2 master = true stats = 127.0.0.1:1717 chdir = /opt/rt4 uid = uitrt gid = uitrt Has anyone else been able to run it using uWSGI? Are there other options for keeping the site running under nginx? Thanks in advance for any pointers or advice. -A Hi Aaron, We run RT with spawn-fcgi/nginx and either systemd/RHEL7 or multiwatch/RHEL6 to make sure that the pool stays populated. No problems at all. I would take uWSGI out of the mix, although there is no reason it should not work, there may be some nuances involved. The fact that the site loads slowly seems to indicate that you are having a problem with your database backend. Check out your slow queries and see what is wrong. It should be pretty zippy by default. Regards, Ken !DSPAM:41,556a90bf10567738710544!
Re: [rt-users] spawn-fcgi crashing, is uWSGI an option?
On Sat, May 30, 2015 at 04:49:00PM -0700, Aaron C. de Bruyn wrote: I have a demo of RT set up for my company. During testing, we found the spawn-fcgi process is slow and crashes often. If more than a few users are on the site, and/or more than few messages coming in through rt-mailgate, the site loads extremely slow. We increased the number of child processes for spawn-fcgi, but it's doesn't seem adequate to handle the load. We can regularly make spawn-fcgi die by simply hitting escape while a page is loading. Other times it crashes for no apparent reason. We use uWSGI everywhere for python projects we host, and I tried to see if I could get RT working under that. I failed miserably because I don't know that much about PSGI, FCGI, etc...and their relationship to the webserver. I know Nginx can speak the uwsgi protocol to talk to uWSGI, but uWSGI seems to bomb on trying to serve RT. I tried the following config to no avail: [uwsgi] plugins = psgi socket = 127.0.0.1:9001 psgi = /opt/rt4/sbin/rt-server.fcgi fastcgi-socket = true processes = 2 master = true stats = 127.0.0.1:1717 chdir = /opt/rt4 uid = uitrt gid = uitrt Has anyone else been able to run it using uWSGI? Are there other options for keeping the site running under nginx? Thanks in advance for any pointers or advice. -A Hi Aaron, We run RT with spawn-fcgi/nginx and either systemd/RHEL7 or multiwatch/RHEL6 to make sure that the pool stays populated. No problems at all. I would take uWSGI out of the mix, although there is no reason it should not work, there may be some nuances involved. The fact that the site loads slowly seems to indicate that you are having a problem with your database backend. Check out your slow queries and see what is wrong. It should be pretty zippy by default. Regards, Ken
[rt-users] spawn-fcgi crashing, is uWSGI an option?
I have a demo of RT set up for my company. During testing, we found the spawn-fcgi process is slow and crashes often. If more than a few users are on the site, and/or more than few messages coming in through rt-mailgate, the site loads extremely slow. We increased the number of child processes for spawn-fcgi, but it's doesn't seem adequate to handle the load. We can regularly make spawn-fcgi die by simply hitting escape while a page is loading. Other times it crashes for no apparent reason. We use uWSGI everywhere for python projects we host, and I tried to see if I could get RT working under that. I failed miserably because I don't know that much about PSGI, FCGI, etc...and their relationship to the webserver. I know Nginx can speak the uwsgi protocol to talk to uWSGI, but uWSGI seems to bomb on trying to serve RT. I tried the following config to no avail: [uwsgi] plugins = psgi socket = 127.0.0.1:9001 psgi = /opt/rt4/sbin/rt-server.fcgi fastcgi-socket = true processes = 2 master = true stats = 127.0.0.1:1717 chdir = /opt/rt4 uid = uitrt gid = uitrt Has anyone else been able to run it using uWSGI? Are there other options for keeping the site running under nginx? Thanks in advance for any pointers or advice. -A
Re: [rt-users] spawn-fcgi crashing, is uWSGI an option?
Thanks Ken. It's working perfectly. -A On Sat, May 30, 2015 at 5:00 PM, k...@rice.edu k...@rice.edu wrote: On Sat, May 30, 2015 at 04:49:00PM -0700, Aaron C. de Bruyn wrote: I have a demo of RT set up for my company. During testing, we found the spawn-fcgi process is slow and crashes often. If more than a few users are on the site, and/or more than few messages coming in through rt-mailgate, the site loads extremely slow. We increased the number of child processes for spawn-fcgi, but it's doesn't seem adequate to handle the load. We can regularly make spawn-fcgi die by simply hitting escape while a page is loading. Other times it crashes for no apparent reason. We use uWSGI everywhere for python projects we host, and I tried to see if I could get RT working under that. I failed miserably because I don't know that much about PSGI, FCGI, etc...and their relationship to the webserver. I know Nginx can speak the uwsgi protocol to talk to uWSGI, but uWSGI seems to bomb on trying to serve RT. I tried the following config to no avail: [uwsgi] plugins = psgi socket = 127.0.0.1:9001 psgi = /opt/rt4/sbin/rt-server.fcgi fastcgi-socket = true processes = 2 master = true stats = 127.0.0.1:1717 chdir = /opt/rt4 uid = uitrt gid = uitrt Has anyone else been able to run it using uWSGI? Are there other options for keeping the site running under nginx? Thanks in advance for any pointers or advice. -A Hi Aaron, We run RT with spawn-fcgi/nginx and either systemd/RHEL7 or multiwatch/RHEL6 to make sure that the pool stays populated. No problems at all. I would take uWSGI out of the mix, although there is no reason it should not work, there may be some nuances involved. The fact that the site loads slowly seems to indicate that you are having a problem with your database backend. Check out your slow queries and see what is wrong. It should be pretty zippy by default. Regards, Ken