Mailing-List: contact users-help@cloudstack.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@cloudstack.apache.org
MIME-Version: 1.0
Date: Thu, 11 Jun 2015 15:04:46 +0100
Message-ID: 
 <CADfcUwGxzGd4poGoOQwV6JnBXu=mUf0VAP93bJWNA37XiYbU-A@mail.gmail.com>
Subject: cloudstack 4.3 - xen hosts - gratuitous ARP
From: cs user <acldstkusr@gmail.com>
To: users@cloudstack.apache.org
Content-Type: multipart/alternative; boundary=94eb2c073bd456b11f05183e792b

--94eb2c073bd456b11f05183e792b
Content-Type: text/plain; charset=UTF-8

Hi Folks,

I'm seeing issues when creating a large number of machines, destroying
them, and quickly recreating them again.... sometimes an ip address which
was previously in use (minutes ago) is reused for the new machines.

We're seeing issues where a switch still believes an enclosure holds the IP
address in question (the old location), whereas in fact it has been
destroyed and started in a different enclosure.

The machine appears to boot as normal and acquires an IP address from the
VR, however further up the stack, the change does not appear to propagate.

One thing to note is, if I console onto the machine, and send some traffic
to the outside world, this does update the arp cache and then I can connect
to the machine from the outside world.

Does the xenhost send a gratuitous arp packet when a new vm is launched on
it?

Once the arp cache timeout is exceeded, everything then works as expected.
So lowering the cache time is one option. Has anyone else had to do this?

Cheers!

--94eb2c073bd456b11f05183e792b--