CharlesH
asked on
Crystal maxing-out CPU
We have Crystal Reports ( 10.0.0.0.5335) and Crystal Enterprise (10.0.0.3885) running on Windows Server 2003 SP2.
Various crystal processes are using up too many processor cycles leading to 100% utilization for days at a time. This leads to the server not responding to requests.
Yes I know that applications use CPU cycles but I am looking for a way to throttle Crystal back so that it leaves some CPU time for the OS. For example, you can tell SQL to only use X GB of memory thereby leaving Y GB of memory for the OS and the other applications.
Is there any way to tell Crystal to never use more than X% of the CPU? Or is there a way to reserve CPU cycles for the OS?
Please note that due to licensing and compatibility issues upgrading or adding CPUs is not an option. (patching is a definite maybe.)
Any other suggestions are most welcome.
Thank you.
Various crystal processes are using up too many processor cycles leading to 100% utilization for days at a time. This leads to the server not responding to requests.
Yes I know that applications use CPU cycles but I am looking for a way to throttle Crystal back so that it leaves some CPU time for the OS. For example, you can tell SQL to only use X GB of memory thereby leaving Y GB of memory for the OS and the other applications.
Is there any way to tell Crystal to never use more than X% of the CPU? Or is there a way to reserve CPU cycles for the OS?
Please note that due to licensing and compatibility issues upgrading or adding CPUs is not an option. (patching is a definite maybe.)
Any other suggestions are most welcome.
Thank you.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
the top users are wmiprvse.exe, JobServerChild.exe, CrystalMS.exe, pageserver.exe and the spooler when they are printing big giant jobs.
CPU = Xeon 3.06 Ghz f2 So my processor is not OVER powered.
I have a request in to see what else is running, where the databases are located and to look at the reports.
I DO know that SQL server 2000 is running on this box as well as Crystal.
I am a network guy and know almost nothing about Crystal.
CPU = Xeon 3.06 Ghz f2 So my processor is not OVER powered.
I have a request in to see what else is running, where the databases are located and to look at the reports.
I DO know that SQL server 2000 is running on this box as well as Crystal.
I am a network guy and know almost nothing about Crystal.
JobServerChild.exe is a process spawned when a scheduled report is running. it might be a good idea to review specific reports to see if they're unnecessarily inefficient (full outer joins, commands or store procedures with outside links or parameters, etc...).
Well,one of the big performance hits that one takes with SQL on older Servers (2003 and below),is a partition alignment issue which can cause a 30 to 40% speed hit when it comes to disk I/O.
Unfortunately the fix is to reinstall the OS from scratch after you have created the partitions and then run the alignment tool.
See:
http://msdn.microsoft.com/en-us/library/dd758814%28v=sql.100%29.aspx
Unfortunately the fix is to reinstall the OS from scratch after you have created the partitions and then run the alignment tool.
See:
http://msdn.microsoft.com/en-us/library/dd758814%28v=sql.100%29.aspx
ASKER
It turns out that the biggest issue was that there were several web-redirects running on this server. In Windows 2003 when you do a web redirect it maintains the connection throughout the sesson. user-redirect server-applicaoion or web server
this is fixed in '08 and later.
So the runaway wmiprvse.exe process was the redirection connections.
I will be tweaking the pageserver timeouts and cache as soon as the server owner comes back from vacation
this is fixed in '08 and later.
So the runaway wmiprvse.exe process was the redirection connections.
I will be tweaking the pageserver timeouts and cache as soon as the server owner comes back from vacation
ASKER
Absolutely wonderful responses all. I awarded points based upon closeness to the biggest issue and for applicability to tuning Crystal.
pgm554, I did not know that about SQL. While SQL is only running on this server to provide ODBC connections to other databases I am going to review the rest of my SQL servers for alignment issues.
(I have already throttled their memory usage and reduced the size of the Windows page files to 6 GB or less.)
pgm554, I did not know that about SQL. While SQL is only running on this server to provide ODBC connections to other databases I am going to review the rest of my SQL servers for alignment issues.
(I have already throttled their memory usage and reduced the size of the Windows page files to 6 GB or less.)
ASKER
I will have to get back to you on most of the specifics.
It is my understanding that there are some other applications that are incompatible with later versions of crystal. We are trying to get away from it as much as possible. We are a SQL shop for the most part.
Pageserver and child processes are one biggie using 30-40% apiece if they can. Also there is a wmi service instance that keeps taking up all available CPU cycles. It has gotten over 60% at times. (It is rare that we are at less than 40% utilization.). That's about all I can come up with from the dentist's.