lholland421 Posted January 19, 2015 Posted January 19, 2015 Hi All, We have just virtualized our file server using Hyper v and all seemed to be running well. We currently have an issue with the servers CPU is currently running up and down all of the time. See the image below. The process using this is "System" . Lately the server has started to drop its connection to users and when it does the CPU idles at 0% - 2%. It brings itself back online again in around 30 seconds. Has anyone else had this issue? Thanks
Michael Posted January 19, 2015 Posted January 19, 2015 This issue has been discussed on Edugeek already in great detail here. The issue appears to be related to the SMB, which 2012 R2 uses 3.x whereas 2008 R2 uses 2.x (top of my head). To the best of my knowledge there isn't a fix.
lholland421 Posted January 19, 2015 Author Posted January 19, 2015 Hope there is a fix as its happening quite a lot at the moment Thanks for the Thread though i will keep my eye on it.
lholland421 Posted January 20, 2015 Author Posted January 20, 2015 Ran updates last night and thought all may have been well but the issue has just started again. I cannot seem to find what these threads are doing within the system Process : Any ideas? They seem to be consuming alot between themselves. Thanks
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 Really thinking about rolling back to 2008 R2.
cpjitservices Posted February 23, 2015 Posted February 23, 2015 Hi @lholland421 - have you installed all the latest updates for 2012 R2 ? I havent come across this issue personally and have a lot of 2012 R2 servers out in the wild... I wouldn't recommend going back to 2008.
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 All the latest updates have been installed, Cannot get my head around it. Its just the system process using ridiculous amounts of CPU. Any other ways to troubleshoot it ?
Koldov Posted February 23, 2015 Posted February 23, 2015 Oh, this is just great news! I just spent the half-term moving all staff shares, home folders and admin drives to the new 2012R2 server... Anyone confirm it is a Hyper-V problem and not something that happens on a 'real hardware' server? Kol.
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 I am wrapping my brains with it. Unreal !
Steve21 Posted February 23, 2015 Posted February 23, 2015 HyperV, 2012R2, none of the problems above. You sure you don't have some over protective AV or something installed doing on access scanning? Steve
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 I have disabled our antivirus and the issue still happens. I have seen people say its related to network drivers but they are up to date
Steve21 Posted February 23, 2015 Posted February 23, 2015 To give you an example in relation to the 2012vs2008 part, this is our staff one running 100~ staff users at 1TB data. Currently using 2.2GB RAM and 2 CPU Steve
kmount Posted February 23, 2015 Posted February 23, 2015 Disclaimer - I'm not a hyper-v expert Hi, I notice in your first screenshot your virtual machine has 10vCPUs and 20GB of RAM allocated to it. I imagine if there's any other VMs running on that host then you will all be fighting for a share of pCPUs and 10vCPU is a LOT for a scheduler to deal with in one chunk. (VMware knowledge talking). May be worth reducing that down to a sensible figure like 2vCPU, same may be relevant for your RAM allocation incase you're forcing paging. As I say, not a hyper-v expert but any VM with 10vCPU is something I'd be wary of.
Steve21 Posted February 23, 2015 Posted February 23, 2015 I have disabled our antivirus and the issue still happens. I have seen people say its related to network drivers but they are up to date Dedup on etc? If so did you disable it during peak times etc Steve
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 I have disable DeDUP but the issue seems to still be there. I have only allocated it 4 cpus with 6 gb ram and this is still happening. When i use the process monitor all the usage is assigned to System. Any other ideas?
kmount Posted February 23, 2015 Posted February 23, 2015 Does it *need* 4? What else is running on the host? What spec is the host? Most of the file servers we look after run happily on 2 and that's with FSRM etc running. The issue with the system process is that it's a bit of a dumping place that can account for usage in many places all adding up. This is an old technet link so not for 2012 but might give some pointers - How to troubleshoot high cpu in the System process - The troubleshooters and problem solvers... - Site Home - TechNet Blogs
Steve21 Posted February 23, 2015 Posted February 23, 2015 Might be easier to know what is and isn't turned on But volume copies? search optimization? Automatic maintenance? Steve
eddyc Posted February 23, 2015 Posted February 23, 2015 We had this issue across the first 2012 R2 file server that we introduced into the mix in that without reason all of the shares would drop. If you log in via the console (in our case through VMware vcentre) you could still log into the domain, still see all of the drives, folders, etc. You just couldn't use the \\servername\share for any reason without a reboot. I spent some time searching online (was considering rolling back to 2008 R2) and some suggested to check that the hosts and drivers are all up to date. I did this out of hours one night and touch wood, six weeks later, the issue seems to have gone away. I think that certain types of NIC in 2012 R2 cause an issue too, they suggest changing the virtual hardware (although I cannot find the article now that I've said that). Hope that might help!
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 I will only assign it 2 cores shortly. Here is the Host Spec
kmount Posted February 23, 2015 Posted February 23, 2015 OT: Just clocked Sophos there, have you got it excluding where your Hyper-V VMs live yeah? The network drivers is a good shout, what nic's are they? How is the storage provided to your Hyper-V environment?
Boredguy Posted February 23, 2015 Posted February 23, 2015 I was always advised that Sophos should be set to a scheduled scan only on servers and not on-Access (which it appears to be as it does not have the yellow warning triangle on it).
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 Shaddow copies is enabled, Automatic maintenance is set to run at 2am in the morning and search optimization is not enabled. Hope this helps
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 The Nics are intel I350 . Sophos is excluded from running where the VM's are stored.
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 The storage is provided from a Dell SAN through 2gb ISCSI connection.
lholland421 Posted February 23, 2015 Author Posted February 23, 2015 I have just disabled the network adaptor and the CPU hit 0% and is fine so it must be NIC related ? Or is it just because no one is trying to browse files?
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now