Hello,
I've recently installed Roundcube 0.2.2 for our users on 1024 RAM with 3Mhz cpu server. But the problem was appearing with high server load. According to the logs we have about 700 messages sent everyday and about 3500 successfull and not sucessfull logins. In the period of all day load is increasing and after midnight it can be about 9 and it caused by Apache processes. That is very strange because in that period almost all users are sleep and we have only few logins and sent messages. This is only one virtual host running on the host. I've tried to use Debian Etch and Lenny.
Aslso I can confirm that after apache restarting high load goes down to 0.2-0.4 but then starts to increase slowly again. So as the temporary solution I use "apache restart" instead "apache reload" in logrotate config.
I guess the problem is in Roundcube somewhere? Dear RC developers. So can you comment this situation? Why it can happen?
Thanks. _______________________________________________ List info: http://lists.roundcube.net/users/
On 5/28/2009 1:31 AM, Andrey wrote:
Hello,
I've recently installed Roundcube 0.2.2 for our users on 1024 RAM with 3Mhz cpu server. But the problem was appearing with high server load. According to the logs we have about 700 messages sent everyday and about 3500 successfull and not sucessfull logins. In the period of all day load is increasing and after midnight it can be about 9 and it caused by Apache processes. That is very strange because in that period almost all users are sleep and we have only few logins and sent messages. This is only one virtual host running on the host. I've tried to use Debian Etch and Lenny.
Aslso I can confirm that after apache restarting high load goes down to 0.2-0.4 but then starts to increase slowly again. So as the temporary solution I use "apache restart" instead "apache reload" in logrotate config.
I guess the problem is in Roundcube somewhere? Dear RC developers. So can you comment this situation? Why it can happen?
Does it only happen at night? If so, maybe some cron job is malfunctioning? Or if it happens all the time, are you certain your server isn't compromised and being used to send out spam through apache?
Hi
I've just encountered a similar performance issue which I have never
noticed on earlier Roundcube versions up to 0.2.1. One single PHP
process was hanging for a looooong time (don't remember, but must have
been 5min+, might as well have been hanging for hours). I was killing
this process and it was gone, no other processes hanging.
here's my current setup (https://webmail.onlime.ch/):
- Roundcube 0.2.2
- Debian Lenny 5.0.1
- Apache 2.2.9
- PHP 5.2.9 (patched with Suhosin 0.9.7 + module 0.9.27 enabled)
- PHP running as CGI under Apache's SuExec
Before killing the process I checked /proc/<PID>/environ to figure out
it was Roundcube. If any Roundcube developer is interested in my
environment variables, I can send you the whole output of /proc/<PID>/
environ off-list. I don't like to post it here for privacy/security
reasons.
Just some information that might help:
HTTP_REFERER=https://webmail.onlime.ch/?_task=mail&_action=preview&_uid=4786&...
REQUEST_URI=/?_task=mail&_action=get&_uid=4786&_mbox=INBOX&_part=49
HTTP_USER_AGENT=Mozilla/5.0 (Macintosh; U; PPC Mac OS X; de-de)
AppleWebKit/523.12.2 (KHTML, like Gecko) Version/3.0.4 Safari/523.12.2
We are talking about a very low volume Roundcube installation, usually
just 5-10 mails sent per day. It's running on a 2x Xeon Quad Core
E5405 2,00GHz / 8GB RAM machine, a powerful server that has a load
average of 0.2
So far, that's the first time I noticed such weird behavior with
Roundcube. I hope it's nothing serious. Good luck in bughunting! thx
Take care Philip
On 28.05.2009, at 07:31, Andrey wrote:
Hello,
I've recently installed Roundcube 0.2.2 for our users on 1024 RAM with 3Mhz cpu server. But the problem was appearing with high server load. According to the logs we have about 700 messages sent everyday and
about 3500 successfull and not sucessfull logins. In the period of all day load is increasing and after midnight it can be about 9 and it
caused by Apache processes. That is very strange because in that period almost
all users are sleep and we have only few logins and sent messages. This is only one virtual host running on the host. I've tried to use Debian
Etch and Lenny.Aslso I can confirm that after apache restarting high load goes down
to 0.2-0.4 but then starts to increase slowly again. So as the temporary solution I use "apache restart" instead "apache reload" in logrotate
config.I guess the problem is in Roundcube somewhere? Dear RC developers. So can you comment this situation? Why it can
happen?Thanks. _______________________________________________ List info: http://lists.roundcube.net/users/
List info: http://lists.roundcube.net/users/
Not that this is your problem, but I found that if someone is logged
into RC, it checks for messages every single minute. RC _does_ need
to keep the connection alive, but it _doesn't_ need to check for
messages that often. So I changed a value in a JavaScript file to
reduce the web server and IMAP server load.
<roundcubedirectory>/program/js/app.js
360c360 < this._int = setInterval(function(){ ref.check_for_recent (); }, this.env.keep_alive * 4900);
this._int = setInterval(function(){ ref.check_for_recent
(); }, this.env.keep_alive * 1000);
The keep alive is still set to the default ( a couple of lines below
the above), but RC won't hit the IMAP server as often.
Earlier in the file this is set :
// default environment vars this.env.keep_alive = 60; // seconds
I'll leave figuring out the actual time(s) in seconds as a math
exercise for any readers.
Charles Marcus пишет:
On 5/28/2009 1:31 AM, Andrey wrote:
Hello,
I've recently installed Roundcube 0.2.2 for our users on 1024 RAM with 3Mhz cpu server. But the problem was appearing with high server load. According to the logs we have about 700 messages sent everyday and about 3500 successfull and not sucessfull logins. In the period of all day load is increasing and after midnight it can be about 9 and it caused by Apache processes. That is very strange because in that period almost all users are sleep and we have only few logins and sent messages. This is only one virtual host running on the host. I've tried to use Debian Etch and Lenny.
Aslso I can confirm that after apache restarting high load goes down to 0.2-0.4 but then starts to increase slowly again. So as the temporary solution I use "apache restart" instead "apache reload" in logrotate config.
I guess the problem is in Roundcube somewhere? Dear RC developers. So can you comment this situation? Why it can happen?
Does it only happen at night? If so, maybe some cron job is malfunctioning? Or if it happens all the time, are you certain your server isn't compromised and being used to send out spam through apache?
Nope, it happens in period of all day. As I can see in RC logs there are no any sent messages that can point to spam through RC. I have the similar system with apache with more than 2500 homepages and more than 500 virtual servers and average load is no more that 1 So I guess it is not OS problem, but RC problem. _______________________________________________ List info: http://lists.roundcube.net/users/
Not that this is your problem, but I found that if someone is logged
into RC, it checks for messages every single minute. RC _does_ need
to keep the connection alive, but it _doesn't_ need to check for
messages that often. So I changed a value in a JavaScript file to
reduce the web server and IMAP server load.<roundcubedirectory>/program/js/app.js
360c360 < this._int = setInterval(function(){ ref.check_for_recent (); }, this.env.keep_alive * 4900);
this._int = setInterval(function(){ ref.check_for_recent(); }, this.env.keep_alive * 1000);
The keep alive is still set to the default ( a couple of lines below
the above), but RC won't hit the IMAP server as often.Earlier in the file this is set :
// default environment vars this.env.keep_alive = 60; // seconds
I'll leave figuring out the actual time(s) in seconds as a math
exercise for any readers.
Thanks for the info. Will try use your advice too. _______________________________________________ List info: http://lists.roundcube.net/users/
Also I've noticed that every time when average server load is growing number of active APACHE a processes is also growing. For example if I do 'netstat -anpl | grep STREAM | grep apache' to get number of active UNIX domain sockets (servers and established) with apache I got 5 apaches when average server load is 5 and 1-2 apaches when load average is quite small.
Then if I grep netstat for PID of these apaches above I see connections to my IMAP server with CLOSE_WAIT state.
Thanks.
Philip Iezzi пишет:
Hi
I've just encountered a similar performance issue which I have never noticed on earlier Roundcube versions up to 0.2.1. One single PHP process was hanging for a looooong time (don't remember, but must have been 5min+, might as well have been hanging for hours). I was killing this process and it was gone, no other processes hanging.
here's my current setup (https://webmail.onlime.ch/):
- Roundcube 0.2.2
- Debian Lenny 5.0.1
- Apache 2.2.9
- PHP 5.2.9 (patched with Suhosin 0.9.7 + module 0.9.27 enabled)
- PHP running as CGI under Apache's SuExec
Before killing the process I checked /proc/<PID>/environ to figure out it was Roundcube. If any Roundcube developer is interested in my environment variables, I can send you the whole output of /proc/<PID>/environ off-list. I don't like to post it here for privacy/security reasons. Just some information that might help:
HTTP_REFERER=https://webmail.onlime.ch/?_task=mail&_action=preview&_uid=4786&...
REQUEST_URI=/?_task=mail&_action=get&_uid=4786&_mbox=INBOX&_part=49 HTTP_USER_AGENT=Mozilla/5.0 (Macintosh; U; PPC Mac OS X; de-de) AppleWebKit/523.12.2 (KHTML, like Gecko) Version/3.0.4 Safari/523.12.2
We are talking about a very low volume Roundcube installation, usually just 5-10 mails sent per day. It's running on a 2x Xeon Quad Core E5405 2,00GHz / 8GB RAM machine, a powerful server that has a load average of 0.2 So far, that's the first time I noticed such weird behavior with Roundcube. I hope it's nothing serious. Good luck in bughunting! thx
Take care Philip
On 28.05.2009, at 07:31, Andrey wrote:
Hello,
I've recently installed Roundcube 0.2.2 for our users on 1024 RAM with 3Mhz cpu server. But the problem was appearing with high server load. According to the logs we have about 700 messages sent everyday and about 3500 successfull and not sucessfull logins. In the period of all day load is increasing and after midnight it can be about 9 and it caused by Apache processes. That is very strange because in that period almost all users are sleep and we have only few logins and sent messages. This is only one virtual host running on the host. I've tried to use Debian Etch and Lenny.
Aslso I can confirm that after apache restarting high load goes down to 0.2-0.4 but then starts to increase slowly again. So as the temporary solution I use "apache restart" instead "apache reload" in logrotate config.
I guess the problem is in Roundcube somewhere? Dear RC developers. So can you comment this situation? Why it can happen?
Thanks. _______________________________________________ List info: http://lists.roundcube.net/users/
List info: http://lists.roundcube.net/users/
-
Andrey -
Charles Marcus -
chasd -
Philip Iezzi