cloudstack-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Amin Samir" <amin.sa...@opencloud.net.au>
Subject RE: Network performance issue
Date Tue, 07 Apr 2015 05:27:31 GMT
Hello,

We had a similar issue that was resolved when we changed the FreeNAS default
sharing behavior from sync to async by adding a new Sysctl with the
following value, 

vfs.nfsd.async   value to 1

-----Original Message-----
From: Sam Ceylani [mailto:sam@mistercertified.com] 
Sent: Friday, April 03, 2015 1:19 AM
To: <users@cloudstack.apache.org>
Subject: Re: Network performance issue

We use Freenas with ZFS also and it is notorious with NFS shares, we created
a dataset and shared on NFS and disabled sync option for this dataset from
ZFS, I forgot the command but it was something like zfs set sync=disabled
/tank/dataset something like that, it will fix it. This has really nothing
to do with cloudstack and xenserver which we use both...


Thanks,

Sam Ceylani, MBA
Computer Engineer
MisterCertified Inc.

301 W. Platt St. Suite 447, Tampa, FL 33606<x-apple-data-detectors://0/0>
P 813<tel:813.264.6460>.264.6460<tel:813.264.6460> M
813<tel:813.416.7867>.416.7867<tel:813.416.7867>
F 800<tel:800.553.9520>.553.9520<tel:800.553.9520> E
sam.ceylani@mistercertified.com<mailto:sam.ceylani@mistercertified.com>

On Apr 2, 2015, at 1:13 PM, "Pankaj Singh"
<pank.sin9191@gmail.com<mailto:pank.sin9191@gmail.com>> wrote:

Still no luck by disabling these flags.
Would you please help us for this.

Thanks & Regards,
PankaJ Singh


On Thu, Apr 2, 2015 at 8:06 PM, Jeronimo Garcia
<garciaj.uk@gmail.com<mailto:garciaj.uk@gmail.com>>
wrote:

mmmmmm i had this problem before ....
i think it was TCP offliading or RX/TX checksumming.
I'd say play with ethtool and see if it makes any difference disabling this
flags (or enabling them) Im on #cloudstack@freenode if you want to pop in.

Thanks

On Thu, Apr 2, 2015 at 3:12 PM, Pankaj Singh
<pank.sin9191@gmail.com<mailto:pank.sin9191@gmail.com>>
wrote:

We have two different network on two similar switches. One is our
192.168.1.0/24 and other is 192.168.2.0/24 network.
Our switch capacity is 1000Mbps and so does our NIC cards on Host machines.

On n/w 192.168.1.0/24, we are using xenserver 6.2 to deploy our Virtual
machines and we can able to achieve persistent 75 to 80 MB/s speed while
transferring (via scp) one 12GiB file from one VM to another VM on same
Xenserver.

We are using 192.168.2.0/24 network for our cloudstack setup.
We have used following software with their version to establish out private
cloud:


Cloudstack = v4.4.2
Management Server On = Centos 6.4 x86_64 bit Xenserver as a hypervisor =
v6.2 (similar configuration of machine as we have used in 192.168.1.0/24
n/w) FreeNAS as a Primary and Secondary Storage via NFS = v9.3 (filesystem
is ZFS  based) Virtual Machine OS = Centos 6.5 x86_64 bit

We are using basic networking with its all default configurations but we are
not able to achieve at least 60 to 65 MB/s of transferring speed.

In cloud If we start scp from one machine to another machine then our
transferring speed starts from 63 MB/s and then its keep falling till 20 to
25 MB/s.

Would you please tell us that what we are missing here. Is ther any kind of
global setting we need to configure so that at least we can achieve more
than 50 MB/s transferring speed between two guest VM.


Thanks and Regards
PankaJ Singh




Mime
View raw message