0

It has weird hangs and then random CPU spikes that do a ton at once. While remoted into the VMs I get an update all at once then it hangs for another 20 seconds. When it lets it go through I get a CPU spike.

Basic server specs for the HW node is: 8 CPUs, 16GB ram 1TB HDD total iPERC6 raid 10

The VMs are barely used but I have them spec'd at VM 1: 4 CPUs, 4GB Ram

VM2: 4 CPUs 6GB ram

The HW node currently says it's total CPU usage is 11% AND Used Memory is at 63%out of 16GB

I'm new to this stuff so I'm not sure. I just recently installed this and set it all up.

UPDATE: I'm still experiencing the same problems, doesn't seem to be as often... TODAY I came into the office, noticed web applications were barely functioning.

I checked the SQL Server VM (windows 2k8 x64) and it was using 90% RAM (well Xen Center showed that). But when I got into the console, it didn't look like it had that much used in task manager.) So I'm thinking Xen Server was so hosed that it wasn't even updating that data regularly or something... who knows...?

How do I get DETAILED logs, when i click logs tab in Xen Center it's like a very basic "start / stop" log file... hmm

UPDATE: (please help) 4/6/2011 I've got it running for a couple weeks in between needing CORE server reboots... but this is still REALLY bad... thse are web servers...

I'm thinking I shouldn't have used 64BIT windows 2008? I'm running ColdFusion 8 32bit, and SQL Server 2005 32bit... could that be causing it?

Improve this question
5
  • Did you install Integration Services on the VMs?
    – DanBig
    Jan 10, 2011 at 15:26
  • I'm not sure... :-\
    – Joshua F. Rountree
    Jan 24, 2011 at 15:31
  • What CPU are you using?
    – DanBig
    Apr 6, 2011 at 13:45
  • We have exactly the same issue, also wiht a Dell PE 2950 III. Did you find a solution? greetz
    – user130808
    Aug 3, 2012 at 7:23
  • To be honest, I don't know a solution... There are various techniques that we have implemented but are unable to determine which actually fixed it. There was a bug in XenServer relating to leaving a mounted disk in the virtual disk drive like an ISO, etc. We dropped our CPU's to 1-2 cpu's per VM and always ensure not to leave an ISO mounted to the main virtual disk drive. That's about all, I'm sure I've ran updates too which has probably resolved it. We no longer have this issue. We also switched most of our VM's to CentOS but I think that is unrelated.
    – Joshua F. Rountree
    Apr 17, 2013 at 12:58

3 Answers 3

Reset to default
2

Unless the VM's actually need that much CPU horsepower, I'd very seriously consider dropping them to one or two processors each. When a VM wants to use a processor cycle, all four cores of one of your physical processors have to be ready, which can cause some pain if anything else is going on (dom0 in this case).

I'm more familiar with the VMware side of things, but the issue is similar. Try going down to at most two cores per VM, or preferably one. Then see how the response time goes.

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1005362

http://communities.vmware.com/thread/143348

Improve this answer
5
  • I think this answer has been helpful, but I'm going to give it a few more days and see how it reacts to the changes. I'll keep you updated.
    – Joshua F. Rountree
    Jan 24, 2011 at 15:32
  • Okay, it's still not helping. They are both only taking up 2 cores out of the total 8 so now a total of 4 cores are in use. It's still hosing up :-\ could it just be XenServer? I REALLY hope there is a fix so that I don't have to redo this server blah :-\
    – Joshua F. Rountree
    Jan 31, 2011 at 13:28
  • Oh, also - it goes through these spurts... it freezes for like... 20 seconds... then all of a sudden lets me do things through console... then it freezes again, keeps doing that... the CPU spikes on the core server
    – Joshua F. Rountree
    Jan 31, 2011 at 13:30
  • hmm... maybe it's my sql server - it's taking up like 3 out of 4 GB of ram for some reason...
    – Joshua F. Rountree
    Jan 31, 2011 at 13:31
  • 1
    Very possible. SQL server will use all available RAM, up to caching all of the database. Make sure you're not overprovisioning your RAM.
    – Jeff McJunkin
    Jan 31, 2011 at 20:32
1

What version of XenServer? If it is 5.6 FP1, install hotfix006 and hotfix007 and retry. About the cpu allocation, you shouldn't allocate more than you have, otherwise you're load the hypervisor scheduler with a huge amount of requests from the VMs. Also, when the VMs are running sluggish, if you run top on dom0, is there any process taking a huge cpu load? How many CPUs are assigned to dom0?

Improve this answer
0

It looks like you may have a memory problem. Install the XenServer guest tools and see if that makes a difference.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .