Discussion in 'LSWS 4.1 Release' started by VortMax, Jul 9, 2009.
"killall lsphp5" set to run at midnight each day.
Thanks... I just implemented something similar albeit on a much more frequent schedule on our server.
You have a PM inbound.
Now I don't think this matters but it won't hurt to try (unless you need to decode php files) is to remove Zend Optimizer temporarily and see if that changes anything. Let's try to isolate the issue and remove any extra possibilities.
Also, can you print screen your external app configuration page on LSWS and attach it to the thread so I can review it?
I have to run Zend for another app that I host.
Can you print screen the configuration details for the lsphp5 app?
Here you go.
Set your max idle time to something between 10 & 30. Set your connection keep alive timeout to something from 30 to 60. Restart. Let me know what happens.
Changes made. Will let you know what happens.
One more question, do you have any other external PHP modules installed? If so, what are they?
Also, when was your PHP 5 built?
One more thing, is this configuration for your main servers external app, or is this for one of your Virtual Host set ups? Are there more than one VH's using the same lsphp5 sockets?
Any fix yet for this problem?
This is killing us too, since we upgraded from 4.0.6 to 4.0.9, lsphp5 processes just accumulate endlessly.
When we do the "killall lsphp5" at command prompt works well.
Wondering do we have to use a cron job before there's a fix?
In our case, we have to clear lsphp5 processes every 20-30 mins, because of a high traffic (5m hits per day) site, which has been working perfectly well before.
1. No other external modules
2. PHP 5.2.9 built May 5, 2009
3. Main server config
The changes that robfrew listed for me above seem to have helped stabilize our server for the past 3 days. Page loads aren't always lightning fast, but our CPU utilization hasn't skyrocketed since making them.
Thanks, I tried & under Server -> External Application ->
Connection Keepalive Timeout = 60
Max Idle Time = 10
But did not help.
I then did was to switch back to version 4.0.6. This helped for sometime
but there're still lsphp5 processes.
One more thing I noticed is this behavior appears to be happening, be'se of the high traffic site. For example, whenever I suspend the high traffic website, all goes back to normal & no more endless lsphp5 processes.
Does this happen with 3.3.24 ?
In my case, just like I mentioned earlier appears to be caused by one high traffic site, and for some strange reason leaves tons of idle lsphp5 processes.
I see this in both versions 4.0.6 and worse when I go to version 4.0.9
How can we resolve this one?
Please download and try 4.0.10 release. just change the version number in the download link.
not sure if this works for your case:
in web admin console define a vhost for this high traffic site, and define a specific listener for it(maybe require an dedicated IP). in listerner setting, binding 1 process only.
The idea is to assign a group of dedicated lsphp5 processes for this high traffic site, isolate with other sites and lsphp5 processes and maybe increase the performance of this site.
Separate names with a comma.