Router creating random

[Keerthiraja SJ]

Hi All,

I am facing a issue whenever I create any VLAN on Advance networking
isolation it create a router where the router creates randomly to the xen
hypervisor.


Some of the xen cluster was dedicated to some other services but when I
create it creates random to all the xen server.

How can I avoid this I want my VR to be create only on particular cluster
group.


Thank,
Keerthi

Re: CloudStack hands out IP address of a stopped VM

[Jayapal Reddy Uradi]

Hi Kyle,

Cloudstack won't delete the nic entry, When VM is deleted it marked as removed.
It might be deleted manually from the DB.

Thanks,
Jayapal


On 11-Apr-2015, at 1:49 AM, Kyle Flavin  wrote:

> Jayapal,
> Is there a reason why would the nic entry be empty?  When is the entry 
> cleared?  It looks like it must have been cleared after the VM was stopped.
> 
> -Original Message-
> From: Jayapal Reddy Uradi [mailto:jayapalreddy.ur...@citrix.com] 
> Sent: Thursday, April 09, 2015 9:01 PM
> To: 
> Subject: Re: CloudStack hands out IP address of a stopped VM
> 
> Hi Kyle,
> 
> When CS picks the ip for new vm from free ip pool excluding nic table 
> ip4_adress.
> In your case the nic entry for vm is empty due to this the ip will goes free 
> pool.
> 
> 
> Thanks,
> Jayapal
> 
> On 10-Apr-2015, at 9:20 AM, Sanjeev N 
> wrote:
> 
>> CS would not hand out the IPs of a stopped vm since the lease time is 
>> infinite. If you are able to reproduce is consistently please open a 
>> JIRA ticket.
>> 
>> On Fri, Apr 10, 2015 at 4:17 AM, Kyle Flavin  wrote:
>> 
>>> Hi Jayapal,
>>> The first query against the nics table returned an empty set, while 
>>> the second returned the hostname and IP address.  Here is the 
>>> sanitized output from both queries:
>>> 
>>> 
>>> mysql> select instance_id,ip4_address  from nics where 
>>> mysql> instance_id=;
>>> Empty set (0.00 sec)
>>> 
>>> 
>>> mysql> select id, name, private_ip_address  from vm_instance  where
>>> id=;
>>> +--+++
>>> | id   | name   | private_ip_address |
>>> +--+++
>>> | myid | myhostname | 1.1.1.1   |
>>> +--+++
>>> 1 row in set (0.00 sec)
>>> 
>>> 
>>> 
>>> -Original Message-
>>> From: Jayapal Reddy Uradi [mailto:jayapalreddy.ur...@citrix.com]
>>> Sent: Thursday, April 09, 2015 5:32 AM
>>> To: 
>>> Subject: Re: CloudStack hands out IP address of a stopped VM
>>> 
>>> Hi Kyle,
>>> 
>>> In my setup I have observed this for stopped VM, the nic table 
>>> ip4_address set to 'null'.
>>> After that I am not able to reproduce the issue.
>>> I will keep looking into my setup for this issue.
>>> 
>>> Can you please send the below commands output from your setup.
>>> 
>>> #select instance_id,ip4_address  from nics where instance_id= 
>>> ; #select id, name, private_ip_address  from vm_instance 
>>> where id=;
>>> 
>>> 
>>> 
>>> Thanks,
>>> Jayapal
>>> 
>>> On 09-Apr-2015, at 6:10 AM, Kyle Flavin  wrote:
>>> 
 I'm trying to get some help understanding the following behavior.
 
 Yesterday we had an instance of CloudStack giving out the IP address 
 of
>>> a stopped VM to a newly created VM.  The existing server was found in 
>>> the MySQL database with the assigned IP (sanitized outputs):
 
 mysql> select name,private_ip_address,state  from vm_instance where 
 mysql> name like "%";
 +++-+
 | name   | private_ip_address | state   |
 +++-+
 |  | 1.1.1.1   | Stopped |
 +++-+
 
 The new server booted up, and was given that same 1.1.1.1 IP as 
 well,
>>> which caused a conflict in our external host management system.
 
 It looks to me like the DHCP lease is expiring on the stopped VM, 
 and
>>> then CloudStack is just handing it out again.  However, it had 
>>> previously been explained to me that CloudStack would not hand out 
>>> IP's of stopped VM's (and I do see the IP address registered to the VM in 
>>> the database).
>>> Is that true and is this a possible bug, or is that the expected behavior?
 
 -Kyle
>>> 
>>> 
> 

ApacheCon

[Remi Bergsma]

Hi all,

If you happen to be at ApacheCon this week, it'd be cool to meet. I’m already 
in Austin with Funs and Daan. Just ping us!

I'll do a talk [1] about automating CloudStack operations that might be of 
interest.

See you there!

Regards,
Remi

[1] http://sched.co/2P8S

Re: Urgent : Xenserver Bond status change

[Fedi Ben Ali]

I think that if this is the case ,ithink that i will have problems to
resolve the issue .But i'm asking if i use a active/passive bond ,can this
help ?

2015-04-12 19:09 GMT+01:00 Sam Ceylani :

> Only explanation I can come up with (and I didn’t want to send this email
> because it could be wrong)
> is since it is using VM's Mac address when registering with your switch
> It might be normal in an active-active bond on public interface, since new
> VM will have a different MAC, or MAC changes on virtual switch
> I might be wrong but I know it uses VM's mac address, like I said it is
> hard to pinpoint the issue but could be the case...
> When I have time I will trace packets on an active-active bond to better
> understand what's going on...
>
> -Original Message-
> From: Fedi Ben Ali [mailto:ben.ali.fe...@gmail.com]
> Sent: Sunday, April 12, 2015 1:59 PM
> To: users@cloudstack.apache.org
> Subject: Re: Urgent : Xenserver Bond status change
>
> Any idea ???
>
> 2015-04-12 0:13 GMT+01:00 Fedi Ben Ali :
>
> > I've cheked my xensource.log and i have no trace of the issue(i keep
> > looking  ,my bond is on Active/Active. All my bonds are configured  on
> > two PIF that are connected to two switches thats why we are not using
> lacp.
> >
> >  Thx
> >
> > 2015-04-11 23:00 GMT+01:00 Sam Ceylani :
> >
> >> adding or removing a vm shouldn't affect management bond status
> >> unless when cloudstack makes an API call to xen toolstack for adding
> >> or removing vm somehow xenserver's management interface is bounced
> >> with that API call, I would check xenserver logs under /var/log and
> >> you can determine the reason, i doubt  it is related to cloudstack,
> >> it could also be a spanning-tree or lacp issue, check switch logs to
> >> determine if any of those sub systems on the switch causing this
> issue...what kind of bond is this?
> >> how is it configured on the switch,is it lacp or not basically?
> >>
> >>
> >> Thanks,
> >>
> >> Sam Ceylani, MBA
> >> Computer Engineer
> >> MisterCertified Inc.
> >>
> >> 301 W. Platt St. Suite 447, Tampa, FL
> >> 33606
> >> P 813.264.6460 M
> >> 813.416.7867
> >> F 800.553.9520 E
> >> sam.ceyl...@mistercertified.com >> m>
> >>
> >> On Apr 11, 2015, at 5:52 PM, "Fedi Ben Ali"  >> > wrote:
> >>
> >> Hello ,
> >>
> >>
> >> I'm working on a platform with many xenservers ,i'm using bonded
> >> interfaces for all the required cloudstack networks
> >> (management,guest,storage,public).
> >>
> >> All my networks works fine except the public network ,each time i add
> >> or remove VM from a cluster ,the xenserver log shows" the status of
> >> the bond has changed ".
> >>
> >> Did i miss some configuration ??
> >>
> >>
> >> i 'll give my log output :
> >>
> >> WB5: Bond status has changed,"3","The status of the eth2+eth3+eth4 bond
> >> changed: 0/3 up (was 3/3)"WB5: Bond status has changed,"3","The
> status
> >> of the eth2+eth3+eth4 bond changed: 3/3 up (was 0/3)"
> >>  WB1: Bond status has changed,"3","The status of the eth5+eth6+eth7
> >> bond
> >> changed: 0/3 up (was 3/3)","
> >>
> >
> >
>

RE: Urgent : Xenserver Bond status change

[Sam Ceylani]

Only explanation I can come up with (and I didn’t want to send this email 
because it could be wrong) 
is since it is using VM's Mac address when registering with your switch 
It might be normal in an active-active bond on public interface, since new VM 
will have a different MAC, or MAC changes on virtual switch
I might be wrong but I know it uses VM's mac address, like I said it is hard to 
pinpoint the issue but could be the case...
When I have time I will trace packets on an active-active bond to better 
understand what's going on...

-Original Message-
From: Fedi Ben Ali [mailto:ben.ali.fe...@gmail.com] 
Sent: Sunday, April 12, 2015 1:59 PM
To: users@cloudstack.apache.org
Subject: Re: Urgent : Xenserver Bond status change

Any idea ???

2015-04-12 0:13 GMT+01:00 Fedi Ben Ali :

> I've cheked my xensource.log and i have no trace of the issue(i keep 
> looking  ,my bond is on Active/Active. All my bonds are configured  on 
> two PIF that are connected to two switches thats why we are not using lacp.
>
>  Thx
>
> 2015-04-11 23:00 GMT+01:00 Sam Ceylani :
>
>> adding or removing a vm shouldn't affect management bond status 
>> unless when cloudstack makes an API call to xen toolstack for adding 
>> or removing vm somehow xenserver's management interface is bounced 
>> with that API call, I would check xenserver logs under /var/log and 
>> you can determine the reason, i doubt  it is related to cloudstack, 
>> it could also be a spanning-tree or lacp issue, check switch logs to 
>> determine if any of those sub systems on the switch causing this 
>> issue...what kind of bond is this?
>> how is it configured on the switch,is it lacp or not basically?
>>
>>
>> Thanks,
>>
>> Sam Ceylani, MBA
>> Computer Engineer
>> MisterCertified Inc.
>>
>> 301 W. Platt St. Suite 447, Tampa, FL 
>> 33606
>> P 813.264.6460 M 
>> 813.416.7867
>> F 800.553.9520 E 
>> sam.ceyl...@mistercertified.com> m>
>>
>> On Apr 11, 2015, at 5:52 PM, "Fedi Ben Ali" > > wrote:
>>
>> Hello ,
>>
>>
>> I'm working on a platform with many xenservers ,i'm using bonded 
>> interfaces for all the required cloudstack networks 
>> (management,guest,storage,public).
>>
>> All my networks works fine except the public network ,each time i add 
>> or remove VM from a cluster ,the xenserver log shows" the status of 
>> the bond has changed ".
>>
>> Did i miss some configuration ??
>>
>>
>> i 'll give my log output :
>>
>> WB5: Bond status has changed,"3","The status of the eth2+eth3+eth4 bond
>> changed: 0/3 up (was 3/3)"WB5: Bond status has changed,"3","The status
>> of the eth2+eth3+eth4 bond changed: 3/3 up (was 0/3)"
>>  WB1: Bond status has changed,"3","The status of the eth5+eth6+eth7 
>> bond
>> changed: 0/3 up (was 3/3)","
>>
>
>

Re: Urgent : Xenserver Bond status change

[Fedi Ben Ali]

Any idea ???

2015-04-12 0:13 GMT+01:00 Fedi Ben Ali :

> I've cheked my xensource.log and i have no trace of the issue(i keep
> looking  ,my bond is on Active/Active. All my bonds are configured  on two
> PIF that are connected to two switches thats why we are not using lacp.
>
>  Thx
>
> 2015-04-11 23:00 GMT+01:00 Sam Ceylani :
>
>> adding or removing a vm shouldn't affect management bond status unless
>> when cloudstack makes an API call to xen toolstack for adding or removing
>> vm somehow xenserver's management interface is bounced with that API call,
>> I would check xenserver logs under /var/log and you can determine the
>> reason, i doubt  it is related to cloudstack, it could also be a
>> spanning-tree or lacp issue, check switch logs to determine if any of those
>> sub systems on the switch causing this issue...what kind of bond is this?
>> how is it configured on the switch,is it lacp or not basically?
>>
>>
>> Thanks,
>>
>> Sam Ceylani, MBA
>> Computer Engineer
>> MisterCertified Inc.
>>
>> 301 W. Platt St. Suite 447, Tampa, FL 33606
>> P 813.264.6460 M
>> 813.416.7867
>> F 800.553.9520 E
>> sam.ceyl...@mistercertified.com
>>
>> On Apr 11, 2015, at 5:52 PM, "Fedi Ben Ali" > > wrote:
>>
>> Hello ,
>>
>>
>> I'm working on a platform with many xenservers ,i'm using bonded
>> interfaces
>> for all the required cloudstack networks
>> (management,guest,storage,public).
>>
>> All my networks works fine except the public network ,each time i add or
>> remove VM from a cluster ,the xenserver log shows" the status of the bond
>> has changed ".
>>
>> Did i miss some configuration ??
>>
>>
>> i 'll give my log output :
>>
>> WB5: Bond status has changed,"3","The status of the eth2+eth3+eth4 bond
>> changed: 0/3 up (was 3/3)"WB5: Bond status has changed,"3","The status
>> of the eth2+eth3+eth4 bond changed: 3/3 up (was 0/3)"
>>  WB1: Bond status has changed,"3","The status of the eth5+eth6+eth7 bond
>> changed: 0/3 up (was 3/3)","
>>
>
>