Microsoft Virtualization Discussions

FAS2020 > 90% CPU



I've just started building VHD files (Hyper-V) but instead of using the Hyper-V GUI each time i've created a 40GB fixed and i'm just copying/pasting to make duplicates. During this process the CPU is sitting at above 90% causing other access to the filer to grind to a hault.

Is this normal for the FAS2020? I have enabled flow control and jumbo frames end-to-end. The host NICs and switch ports have been set to 1000Mbps FD but I understand the FAS2020 cannot be specifically set to 1000Mbps FD so it's auto neg.

Using Microsoft iSCSI Initiator on 2008 R2 with 2 dedicated NICs and switches to the FAS2020. TOE etc on the host NICs are set to TX/RX.

Any ideas?



What is your Network IOPS and Network input/outpu in "Filer At-A-Glance" tool?

//Thomas Spindler



Haven't check At-a-Glance yet but System Manager the I/O (bytes/sec) is peaking at about 0.4M, the Operations (ops/sec) is peaking at about 1K and the Read Latency (msec) is peaking at 8.0. If that helps.

I will check out At-a-Glance now as well.



To me it looks like you are maxing out the filer, as your copy/paste operation actually means you are reading 40GB file from the disk & then writing it back, whilst every I/O goes via Hyper-V host.

Personally I'd rather do either FlexClone (if you have a license) of my golden image, or LUN clone (could be even better, as you can keep multiple LUNs in the same volume).

After you are done with the desired number of VMs, you can split the clones - this operation will generate a lot of disk I/O too, but at least no traffic will be traversed via host.




That would be rather worrying if a bit of equipment in the region of 10 of thousands of pounds is being maxed out by simply copying and pasting a file (of any size).

I am now in the process of creating a 250GB VHD (for file server data) and the CPU is maxing out again. There is no reading going on here as it's been done within the Hyper-V GUI this time.


I don't really want the same number of LUNs as I do VMs, otherwise that would mean almost 15 LUNs. I am using 2008 R2 with Cluster Shared Volumes.


I'd just like to state that I am also having an issue with my FAS2020 spiking to 90% CPU when doing a simple copy of a file from one folder to another inside an iSCSI LUN attached to a Vsphere VM.

I don't think there is a Netapp tool which will do this for me.



1st point the sysstat -m shows the right utilization of the CPUs and when you do the sysstat it show CPU utilization more when more than 1 CPU is being used at good rate , it does not mean that your CPU is highly utilized. Checked it out by sysstat -m

2nd word what you want to do, do you want to make the clone of files , ok if this is your query than please you can have the vol clone command clone the volume and then you can target that cloned volume where you want. But take care that cloned volume is created using the snapshot so if the snapshot is being carried out via script then script will fail at that time if that snap used for creating the LUN is still in use and that leads to the replication failure if that happen. you can split the clone volume do remember you should have space in the aggregate to guarantee the space to new volume being split.

The FAS2020 is a single CPU/single core system.  Running "sysstat -m" will give you an unrecognized command response.