Solved

Poor performance on a new Server 2008 R2 Standard print server.

Posted on 2013-12-19
17
759 Views
Last Modified: 2014-06-12
Hello,

We have recently sunsetted our 2003 print servers.  We were transitioning to a managed-print vendor who recommended a very different print solution, so rather than migrate, we created two new print servers.  The drivers are current (provided by the MPS provider).

The problem print server is Server 2008 R2 Standard, running virtually on a VMware platform.  It's configured with 12GB Ram and 4 processor cores (2 virtual sockets, 2 cores per socket).

It has the file server and print server roles installed (although the file server role is NOT being utilized).

There are currently 178 printers configured on the server.  Jobs are set to render on the client computers.  We deploy the printers using group policy (AD 2008).

Printing seems largely ok except that the processor utilization spikes and stays high for 5-10 minutes at a time many times during the day.  'High' utilization varies from 75% to 100%.  'Resting' (no active print jobs) cpu usage during school hours stays fairly steady at around 70%.  Off hours the CPU utilization hovers around 40%.

Printing seems ok, but we are getting sporadic reports of printers 'disappearing' from workstations, then reappearing on their own (or after a reboot).  We've also had cases where the printers weren't showing up on a workstation upon logon, only to appear a few minutes later.  I attributed these at first to some issue with group policy, but I'm now wondering if maybe we aren't seeing the product of the server cpu being pegged and the printer objects on the workstations cannot be rendered properly during that time period?

I'm at a loss right now to explain what is happening.  I'd expect to see spikes during the actual print jobs, but I'd also expect to see the spike drop back down to normal levels relatively quickly.


Thanks for any input you might have!

Scott
0
Comment
Question by:meelnah
  • 7
  • 5
  • 3
  • +1
17 Comments
 
LVL 119

Accepted Solution

by:
Andrew Hancock (VMware vExpert / EE MVE^2) earned 250 total points
ID: 39730344
Have you evaluated the need for 4 vCPUs?

e.g. have you tried checking if the issues, disappear, by reducing to just 2 vCPU (2 sockets, 1 core).
0
 
LVL 22

Assisted Solution

by:Haresh Nikumbh
Haresh Nikumbh earned 250 total points
ID: 39730354
The problem print server is Server 2008 R2 Standard, running virtually on a VMware platform.  It's configured with 12GB Ram and 4 processor cores (2 virtual sockets, 2 cores per socket).

change it to Number of virtual Sockets : 2
Number of Cores per Virtual Socket :1
0
 
LVL 1

Author Comment

by:meelnah
ID: 39730359
So you both are saying that we are using too many cpus?

We started with a base installation (I believe) of 1 cpu socket, 1 core (I could be wrong here).  When the cpu utilization issues became apparent, we began adding processors to attempt to get a handle on it.

Are you both thinking that we have given the print server too much cpu?
0
Simplifying Server Workload Migrations

This use case outlines the migration challenges that organizations face and how the Acronis AnyData Engine supports physical-to-physical (P2P), physical-to-virtual (P2V), virtual to physical (V2P), and cross-virtual (V2V) migration scenarios to address these challenges.

 
LVL 119
ID: 39730371
It's possible.....but if you started with a single processor, 1 vCPU (socket) 1 (core), and it was performing poorly, and you added another and it was still poor....

but I would only use cores, for licensing reasons.

so I would try....2 vCPUs (2 sockets), 1 (core).

But, what is the storage system? e.g. VMFS datastore? SATA 7k disks?
0
 
LVL 22

Expert Comment

by:Haresh Nikumbh
ID: 39730372
yep adding CPU create Overhead on the host and then it increase CPU  ready time which is one the reason affect VM CP performance

https://blogs.vmware.com/vsphere/2013/10/does-corespersocket-affect-performance.html
0
 
LVL 22

Expert Comment

by:Haresh Nikumbh
ID: 39730374
If this configuration does not help then need to check on the storage bottleneck + or network utilization chart
0
 
LVL 119
ID: 39730388
As I've just read this is a school environment, we recommend to our schools, and we configure, the print spoolers to restart daily! (at 7.00am) before the day begins.

CPU can be an issue.
0
 
LVL 1

Author Comment

by:meelnah
ID: 39730423
~hanccocka ... we are using an XIO ISE2 SAN for storage (10k SAS drives), so I don't think the bottleneck is memory.

~takecoffe ... thanks for the article.  That's an interesting read, and I'm going to try reconfiguring the processors tonight to see if that helps.

~hanccocka ... I like the idea of restarting the spooler.  I just set a job to restart it daily.


To you both... 178 printers on a single server doesn't seem to be an issue then?  In my research it didn't seem like there was a standard answer to 'how many printers per server?'  I'm sure it has more to do with the number of print jobs than defined printers.
0
 
LVL 119
ID: 39730449
We have got 480 printer queues on a clustered print server.

Never had issues and that only has single processors.

BUT we do find on occasion a corrupt print driver can cause issues or bad print driver we try to concentrate on HP printers if we can and Postscript or PCL6 drivers

Some drivers for home printers are rubbish not designed for heavy use.

So be warned if you have got a right mixup of cheap printers....

Oh yes we have been their with many schools and scrapped all their printers and netgear or belkin print servers
0
 
LVL 1

Author Comment

by:meelnah
ID: 39730455
When you say clustered print server, is it a failover type cluster, or load balancing?

We've had the issues with a corrupt driver as well.  We have a pretty rigid printer selection process now (3 vendors, 9 total types throughout our district), so we've had much better success with the drivers.
0
 
LVL 119
ID: 39730470
Microsoft Failover Cluster.
0
 
LVL 1

Author Comment

by:meelnah
ID: 39730484
Thanks to you both.

I'll update the ticket tomorrow with the results... i'll plan on moving to 2 sockets, 1 processor per socket (per the article).

Scott
0
 
LVL 1

Author Comment

by:meelnah
ID: 39731844
Hi,

I changed the virtual machine to 2 cpu sockets, 1 processor per socket, and this morning as users logged on the cpu pegged at 100% for 20 minutes -- people who were logged on couldn't get their print jobs to go, people who weren't logged on didn't get their printers mapped when they logged on to the network.

Powered down the print server and moved to four processor sockets, 1 core per socket.  It's printing now, but the cpu is still running at around 80% utilization in perfmon.  Spoolersv is at around 55%.

Our managed print vendor recommended rendering the print jobs on the clients rather than the server.  Is this how yours are set up?

Another question that comes up is whether we should be setting processor affinity for the spooler service, or possibly prioritizing the spooler service in task manager... do either of you have experience there?

Thanks,

Scott
0
 
LVL 19

Expert Comment

by:compdigit44
ID: 39740936
What type of documents are your users printing. I ran into an odd issue about two years ago where our printer server kept crash, come to find out the app users where printing from was actually opening each document in MS Word on the print server before printing it!!!!!

So you could easily see how 20 or 30 print job with MS WOrd open on a server would make it crash.

Have you tried to run Process Explorer on the printer server turning period of high volume printing?

ON you print server are you using universal drivers and/or print driver isolation?
0
 
LVL 1

Author Comment

by:meelnah
ID: 39742180
Hi Compdigit,

There's no real pattern to what the users are printing.  It's a fairly large school district, so we have word docs (Office 2013), google docs, jpgs, spreadsheets, etc.  I don't have office installed on the print server, so no issues with the documents trying to be opened on the print server to print.

Print jobs are set to render on the client workstation, not on the server.

Print driver isolation is enabled.

During the high cpu utilization times, I typically see the print spooler service run up to 70 -80%.  Other running processes (system, VMware, crss, etc. all run normally).
0
 
LVL 19

Expert Comment

by:compdigit44
ID: 39742397
I believe another experts mentioned this earlier about corrupt print drivers.  I would suggest reading the following article on how to clear a corrupt driver..

http://yourcomputer.in/print-spooler-spoolsv-exe-causing-high-cpu-usage/

Good Luck
0
 
LVL 1

Author Closing Comment

by:meelnah
ID: 40130360
Sorry about forgetting to close the question.  The problem persists, but I'm closing the question and awarding points for helping me identify improperly configured CPUs in VMWare.
0

Featured Post

Best Practices: Disaster Recovery Testing

Besides backup, any IT division should have a disaster recovery plan. You will find a few tips below relating to the development of such a plan and to what issues one should pay special attention in the course of backup planning.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

If we need to check who deleted a Virtual Machine from our vCenter. Looking this task in logs can be painful and spend lot of time, so the best way to check this is in the vCenter DB. Just connect to vCenter DB(default DB should be VCDB and using…
In this article, I will show you HOW TO: Perform a Physical to Virtual (P2V) Conversion the easy way from a computer backup (image).
This tutorial will show how to configure a single USB drive with a separate folder for each day of the week. This will allow each of the backups to be kept separate preventing the previous day’s backup from being overwritten. The USB drive must be s…
This tutorial will walk an individual through setting the global and backup job media overwrite and protection periods in Backup Exec 2012. Log onto the Backup Exec Central Administration Server. Examine the services. If all or most of them are stop…

839 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question