Help, desparately strange sugarcrm speed problem?

[Gill]

I have been trialling a sugarcrm system. Version 4.2.1a installed to a Windows XP SP2 machine using the bitrock installer.

Yesterday I was demoing the system to my boss and everything was running at near instance access speed.

Today I have a terrible speed issue.

When moving to a new screen (eg switching from the contacts tab to the accounts tab) takes 10+ seconds, sometimes it times out completely. If it does not time out then the server response speed at the bottom of the screen always indicates it was less than 1 second response time. This speed issue only effects machines connecting to it over the network, the local machine accesses Sugarcrm fine.

Summary
Machine 1 NEW 2.68ghz pentium D processor 1 gb ram (server computer with problem), SATA HDD
Machine 2 AMD 2ghz (server computer without problem) 512mb ram, IDE HDD


Machine 1 Connects to Machine 2 - normal speed
Machine 1 connects to machine 1 - normal speed
Machine 2 connects to machine 1 - very slow*
Machine 2 connects to machine 2 - normal speed

*note i have tried 4 machines all so far have problems accessing machine 1, so the problem is not machine 2 or browser cache...

I have looked through the apache log file and cannot see any problems, also running other websites from machine 2 on machine 1, such as Mambo, or phpmyadmin are all fine.

To my mind

-cannot be a network issue as speed tests transferring files show no problems
-cannot be apache as other websites using php and mysl are fine
-cannot be a corruption of sugarcrm as local access to sugarcrm is fine
-cannot be browser cache as I have cleared this and tried several different machines.
-cannot be apache cache as I have restarted the server
-cannot be user specific as I am logged on as the same user.

If anyone can supply any ideas as to why sugar is so slow I would love to here them...

[Gill]

Please note I have tested te above problem in IE, Firefox and Opera, all browsers exhibit the strange behaviour above.

There have been no code base changes since the problem first occurred (or for several weeks). The only change to the database since the problem has been the addition of a mail campaign (which has not been fully completed) as I was testing this functionality when the problem first occurred. The exact time the problem started was when I created an email template. When I have access to the system again tomorrow I will see if I can cause the problem again on the other system.

The two installed versions I have are both 4.2.1a and are identical except the one with slow down has additional test data in the database.

There appears to be no problem with the sugarcrm database tables as I have been able to construct a basic php page that queries and lists all the data in the tables.

If anyone has any suggestions of where to look next I would be greatful.

Thanks.

[kpit]

Have you look to see if you have a memory problem? something that just kills a machines performance is running out of real memory and using swap space as memory. A few key indicators is your harddrive access light is always on. SugarCRM is somewhat a memory pig when you have many people accessing the system at once.

Another thing to check is other programs or processes running on that server that are consuming resources such as memory or processor time. For example an email server running on the same machine?

Another cause for delayed load is DNS lookups to find the site. Then it connects to the server after it finds the page.

Turn on Server response times in the admin->System Settings. Look at the response times at the bottom of the page you just loaded. The times should be arround 1 or 2 seconds. ( I am using a 3Ghz Xeon with hardware raid mine may be faster). This displays how much server time it is taking to generate the page.


These are the most common problems.

[Gill]

Have you look to see if you have a memory problem? Another thing to check is other programs or processes running on that server that are consuming resources such as memory or processor time. For example an email server running on the same machine?

650mb physical memory free and 3% cpu use when problem is occurring. No other processes are running on this machine other than normal windows ones, even the firewall and auto update are turned off.

Another cause for delayed load is DNS lookups to find the site. Then it connects to the server after it finds the page.

I am connecting via IP address. If finds the IP and Firefox reports that it is "Read IP 192....."

Turn on Server response times in the admin->System Settings. Look at the response times at the bottom of the page you just loaded. The times should be arround 1 or 2 seconds. ( I am using a 3Ghz Xeon with hardware raid mine may be faster). This displays how much server time it is taking to generate the page.

This is always less than 1 second, The page is taking 15-20 seconds to display.

These are the most common problems.

OK I am 1 step closer to solving the problem.

1. Connect to SugarCRM, click on a few tabs, the first few tabs are sluggish, but after that the third or more often time out, from this point on any navigation is very slow.
2. Connect to Mambo on the same server, although SugarCRM is unusable mambo is still running fine, but after a few clicks to navigate around it is unusable.
3. Connect to Joomla on the same server, although now both Mambo and SugarCRM are unusable Joomla works, until a few navigation clicks and you cannot use Joolma.
4. During all of this the server CPU (0-1%), Network utilisation (0-2%) and Memory (50% - does not change) are not taxed in the slightest.
5. Connect directly from the server to these applications and all are working fine, and you cannot break them by navigating, the remote connection however is still not working.
6. Use the UNC network path to connect to the server and browse to the htdocs\sugarcrm, and I notice that sugarcrm takes a very long time to read the data, tried to copy a file and this timed out....

-----
As I was typing this I solved the problem. Zend Platform was killing the remote connection. I do not know the exact details of the problem but after uninstalling this the system now runs as expected. I am going to investigate, perhaps it was some settings that were very wrong, essentially I installed the latest Zend Studio on the machine two weeks ago and it did not present any problems until yesterday.

My guess was that the optimizer was creating some sort of queue delay on remote connections with the assumption that thousands of people were connecting...what ever it was doing it was doing this very badly.