Hi!
So i figured out that it was due to MTU settings.
In order for the guest traffic on host`s physical NIC eth0.900 to traverse the
switch I had to reduce the MTU on the guest VM to 1496.
My guess is because the vlan tag 900 adds 4 bytes to the packets sent out the
host`s NIC eth0.900. My switch is not configured for fragmentation so the
packets where being dropped.
On Tuesday, November 12, 2013 1:35 AM, Andrei Mikhailovsky and...@arhont.com
wrote:
Jake, what network card are you using? Is it a broadcom chip by any chance?
Andrei
- Original Message -
From: Jake G. dj_dark_jungl...@yahoo.com
To: users@cloudstack.apache.org
Sent: Monday, 11 November, 2013 4:44:16 AM
Subject: Networking issues when vRouter is on different host
Hi All,
I'm running Cloudstack 4.2 on CentOS 6.4 with 2 KVM hosts using advanced
networking.
I have 3 networks on 2 physical NICs.
eth0 - Management, Storage traffic
eth0.900 - Guest Network VLAN 900 on the physical NIC, but a Guest VM vlan
range of 600-699 handled by the CS vRouter.
eth1 - Internet, public traffic
I am able to deploy Guest VMs in the cluster containing the 2 KVM hosts. I can
also ping the vRouter and other Guest VMs located on the other host of the
cluster. Lastly, I can receive Guest IPs from the vRouter`s DHCP service no
problem.
My problem is, I am unable to connect to the internet and/or the internet is
extremely slow (200bytes/sec) from a Guest VM that is located on a different
host than the vRouter. Guest machines that are located on the same host of the
vRouter work fine.
Could this possibly be a MTU issue? As far as I know everything should be the
default MTU 1500.
Thank you very much!