The host node has 96GB ram and 2 E5-2620 cpus.There is just one vm for test on this node. The storage node is just only working for the host node.
I don't know if cloudstack has some default resource limit policies. I don't think workload is the reason.There may be a system feature which cause this issue,about cloudstack,or kvm,or windows system,or storage device,or something else. Does any one have relevant experience in dealing with this issue? ------------------ Original ------------------ From: "David Comerford"<davest...@gmail.com>; Date: Sun, Jul 7, 2013 05:43 AM To: "users"<users@cloudstack.apache.org>; Subject: Re: vm's disk io performance declines quickly after being created0.5-1 hours later. Doesn't sounds like a CloudStack problem. Spin up a Linux instance and see does it have the same problem. Then apply some science and check I/O inside the instance. Then I/O on the primary storage device. Keep an eye on the storage NICs too. Does you host have enough RAM? Maybe it's booting up in main memory and starting to swap over the hour. Best regards, David Comerford Mobile: +353 87 1238295 Email: davest...@gmail.com Website: http://dave.ie GPG key: http://pgp.dave.ie On 6 July 2013 10:56, WXR <1485739...@qq.com> wrote: > I create a windows 2003 server enterprise 32bit vm template. > When I use this template to create a vm instance and start the vm.The disk > io performance is normal,it just takes 5 seconds to copy a 200MB file.And > other files/directories copy jobs are very fask. > > But 0.5-1 hours later,the vm need serveral minutes to copy a 200MB > file,and all other files copy job become very slow. > > I don't know if it is an issue of windows 2003 vm template,or cloudstack > configuration,or storage configuration. > Does anyone meet the same issue?
