You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@cloudstack.apache.org by Jeronimo Garcia <ga...@gmail.com> on 2013/04/10 18:27:09 UTC
System VMs Failover
Hi List.
Based on documentation and what I'm actually seeing now , system vms can
only failover to other hosts in the same cluster and therefore pod.
I'm guessing this is because the system vm is using a management ip
assigned to that specific pod.
Is there any HA offering that could make system VMs to failover to
different pods/clusters?
Thanks!
Re: System VMs Failover
Posted by Ahmad Emneina <ae...@gmail.com>.
it could be per cluster or per host (local storage) at the moment. I
believe after the storage refactor work happens is when support for zone
wide primary *might* become a reality.
On Thu, Apr 11, 2013 at 7:28 PM, Jeronimo Garcia <ga...@gmail.com>wrote:
> Thanks.
>
> Is primary storage zone wide too?
>
>
>
> On Thu, Apr 11, 2013 at 9:06 PM, Chiradeep Vittal <
> Chiradeep.Vittal@citrix.com> wrote:
>
> > Secondary storage is zone wide not pod-specific.
> > Pod2 probably has some problems (enough shared storage perhaps?). The MS
> > logs should indicate the problem.
> >
> > On 4/11/13 7:03 PM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
> >
> > >just to clarify I'm using KVM.
> > >
> > >Thanks!
> > >
> > >
> > >On Thu, Apr 11, 2013 at 1:26 PM, Jeronimo Garcia
> > ><ga...@gmail.com>wrote:
> > >
> > >> Hi.
> > >>
> > >> Thanks for your answer .
> > >> System vms images are normally in secondary storage , and currently i
> > >> have 3 pods and one and only secondary storage nfs system,
> > >>
> > >> When i shut down all the servers on POD1 , the system vms refused to
> > >> launch in pod2 , would it be cause Pod2 has to have it's own
> secondary
> > >> storage.
> > >>
> > >> Thanks!
> > >>
> > >>
> > >> On Thu, Apr 11, 2013 at 12:08 PM, Kelven Yang
> > >><ke...@citrix.com>wrote:
> > >>
> > >>> System VMs are tend to be running in stateless mode, instead of
> > >>>fail-over
> > >>> of same VM, we are launching new instance of them, this behave is
> done
> > >>>for
> > >>> storage and console proxy system VM, for virtual router system VM,
> > >>>there
> > >>> is a redundant configuration, automatic HA would make the redundant
> > >>> configuration unstable. so in general, system VMs don't use HA
> offering
> > >>> purposely.
> > >>>
> > >>> Kelven
> > >>>
> > >>> On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
> > >>>
> > >>> >Anyone ever bumped into this yet?
> > >>> >
> > >>> >Thanks!
> > >>> >
> > >>> >
> > >>> >On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
> > >>> ><ga...@gmail.com>wrote:
> > >>> >
> > >>> >> Hi List.
> > >>> >>
> > >>> >> Based on documentation and what I'm actually seeing now , system
> > >>>vms
> > >>> >>can
> > >>> >> only failover to other hosts in the same cluster and therefore
> pod.
> > >>> >>
> > >>> >> I'm guessing this is because the system vm is using a management
> ip
> > >>> >> assigned to that specific pod.
> > >>> >>
> > >>> >> Is there any HA offering that could make system VMs to failover to
> > >>> >> different pods/clusters?
> > >>> >>
> > >>> >> Thanks!
> > >>> >>
> > >>>
> > >>>
> > >>
> >
> >
>
Re: System VMs Failover
Posted by Jeronimo Garcia <ga...@gmail.com>.
Thanks.
Is primary storage zone wide too?
On Thu, Apr 11, 2013 at 9:06 PM, Chiradeep Vittal <
Chiradeep.Vittal@citrix.com> wrote:
> Secondary storage is zone wide not pod-specific.
> Pod2 probably has some problems (enough shared storage perhaps?). The MS
> logs should indicate the problem.
>
> On 4/11/13 7:03 PM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>
> >just to clarify I'm using KVM.
> >
> >Thanks!
> >
> >
> >On Thu, Apr 11, 2013 at 1:26 PM, Jeronimo Garcia
> ><ga...@gmail.com>wrote:
> >
> >> Hi.
> >>
> >> Thanks for your answer .
> >> System vms images are normally in secondary storage , and currently i
> >> have 3 pods and one and only secondary storage nfs system,
> >>
> >> When i shut down all the servers on POD1 , the system vms refused to
> >> launch in pod2 , would it be cause Pod2 has to have it's own secondary
> >> storage.
> >>
> >> Thanks!
> >>
> >>
> >> On Thu, Apr 11, 2013 at 12:08 PM, Kelven Yang
> >><ke...@citrix.com>wrote:
> >>
> >>> System VMs are tend to be running in stateless mode, instead of
> >>>fail-over
> >>> of same VM, we are launching new instance of them, this behave is done
> >>>for
> >>> storage and console proxy system VM, for virtual router system VM,
> >>>there
> >>> is a redundant configuration, automatic HA would make the redundant
> >>> configuration unstable. so in general, system VMs don't use HA offering
> >>> purposely.
> >>>
> >>> Kelven
> >>>
> >>> On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
> >>>
> >>> >Anyone ever bumped into this yet?
> >>> >
> >>> >Thanks!
> >>> >
> >>> >
> >>> >On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
> >>> ><ga...@gmail.com>wrote:
> >>> >
> >>> >> Hi List.
> >>> >>
> >>> >> Based on documentation and what I'm actually seeing now , system
> >>>vms
> >>> >>can
> >>> >> only failover to other hosts in the same cluster and therefore pod.
> >>> >>
> >>> >> I'm guessing this is because the system vm is using a management ip
> >>> >> assigned to that specific pod.
> >>> >>
> >>> >> Is there any HA offering that could make system VMs to failover to
> >>> >> different pods/clusters?
> >>> >>
> >>> >> Thanks!
> >>> >>
> >>>
> >>>
> >>
>
>
Re: System VMs Failover
Posted by Chiradeep Vittal <Ch...@citrix.com>.
Secondary storage is zone wide not pod-specific.
Pod2 probably has some problems (enough shared storage perhaps?). The MS
logs should indicate the problem.
On 4/11/13 7:03 PM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>just to clarify I'm using KVM.
>
>Thanks!
>
>
>On Thu, Apr 11, 2013 at 1:26 PM, Jeronimo Garcia
><ga...@gmail.com>wrote:
>
>> Hi.
>>
>> Thanks for your answer .
>> System vms images are normally in secondary storage , and currently i
>> have 3 pods and one and only secondary storage nfs system,
>>
>> When i shut down all the servers on POD1 , the system vms refused to
>> launch in pod2 , would it be cause Pod2 has to have it's own secondary
>> storage.
>>
>> Thanks!
>>
>>
>> On Thu, Apr 11, 2013 at 12:08 PM, Kelven Yang
>><ke...@citrix.com>wrote:
>>
>>> System VMs are tend to be running in stateless mode, instead of
>>>fail-over
>>> of same VM, we are launching new instance of them, this behave is done
>>>for
>>> storage and console proxy system VM, for virtual router system VM,
>>>there
>>> is a redundant configuration, automatic HA would make the redundant
>>> configuration unstable. so in general, system VMs don't use HA offering
>>> purposely.
>>>
>>> Kelven
>>>
>>> On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>>>
>>> >Anyone ever bumped into this yet?
>>> >
>>> >Thanks!
>>> >
>>> >
>>> >On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
>>> ><ga...@gmail.com>wrote:
>>> >
>>> >> Hi List.
>>> >>
>>> >> Based on documentation and what I'm actually seeing now , system
>>>vms
>>> >>can
>>> >> only failover to other hosts in the same cluster and therefore pod.
>>> >>
>>> >> I'm guessing this is because the system vm is using a management ip
>>> >> assigned to that specific pod.
>>> >>
>>> >> Is there any HA offering that could make system VMs to failover to
>>> >> different pods/clusters?
>>> >>
>>> >> Thanks!
>>> >>
>>>
>>>
>>
Re: System VMs Failover
Posted by Jeronimo Garcia <ga...@gmail.com>.
just to clarify I'm using KVM.
Thanks!
On Thu, Apr 11, 2013 at 1:26 PM, Jeronimo Garcia <ga...@gmail.com>wrote:
> Hi.
>
> Thanks for your answer .
> System vms images are normally in secondary storage , and currently i
> have 3 pods and one and only secondary storage nfs system,
>
> When i shut down all the servers on POD1 , the system vms refused to
> launch in pod2 , would it be cause Pod2 has to have it's own secondary
> storage.
>
> Thanks!
>
>
> On Thu, Apr 11, 2013 at 12:08 PM, Kelven Yang <ke...@citrix.com>wrote:
>
>> System VMs are tend to be running in stateless mode, instead of fail-over
>> of same VM, we are launching new instance of them, this behave is done for
>> storage and console proxy system VM, for virtual router system VM, there
>> is a redundant configuration, automatic HA would make the redundant
>> configuration unstable. so in general, system VMs don't use HA offering
>> purposely.
>>
>> Kelven
>>
>> On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>>
>> >Anyone ever bumped into this yet?
>> >
>> >Thanks!
>> >
>> >
>> >On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
>> ><ga...@gmail.com>wrote:
>> >
>> >> Hi List.
>> >>
>> >> Based on documentation and what I'm actually seeing now , system vms
>> >>can
>> >> only failover to other hosts in the same cluster and therefore pod.
>> >>
>> >> I'm guessing this is because the system vm is using a management ip
>> >> assigned to that specific pod.
>> >>
>> >> Is there any HA offering that could make system VMs to failover to
>> >> different pods/clusters?
>> >>
>> >> Thanks!
>> >>
>>
>>
>
Re: System VMs Failover
Posted by Jeronimo Garcia <ga...@gmail.com>.
Hi.
Thanks for your answer .
System vms images are normally in secondary storage , and currently i have
3 pods and one and only secondary storage nfs system,
When i shut down all the servers on POD1 , the system vms refused to
launch in pod2 , would it be cause Pod2 has to have it's own secondary
storage.
Thanks!
On Thu, Apr 11, 2013 at 12:08 PM, Kelven Yang <ke...@citrix.com>wrote:
> System VMs are tend to be running in stateless mode, instead of fail-over
> of same VM, we are launching new instance of them, this behave is done for
> storage and console proxy system VM, for virtual router system VM, there
> is a redundant configuration, automatic HA would make the redundant
> configuration unstable. so in general, system VMs don't use HA offering
> purposely.
>
> Kelven
>
> On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>
> >Anyone ever bumped into this yet?
> >
> >Thanks!
> >
> >
> >On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
> ><ga...@gmail.com>wrote:
> >
> >> Hi List.
> >>
> >> Based on documentation and what I'm actually seeing now , system vms
> >>can
> >> only failover to other hosts in the same cluster and therefore pod.
> >>
> >> I'm guessing this is because the system vm is using a management ip
> >> assigned to that specific pod.
> >>
> >> Is there any HA offering that could make system VMs to failover to
> >> different pods/clusters?
> >>
> >> Thanks!
> >>
>
>
Re: System VMs Failover
Posted by Chiradeep Vittal <Ch...@citrix.com>.
Is this on vSphere?
For other hypervisors, theoretically HA can start it on other clusters
(since the control ip is link local). But if there is capacity available
on the first cluster, why not restart it on the original cluster?
On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>Anyone ever bumped into this yet?
>
>Thanks!
>
>
>On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
><ga...@gmail.com>wrote:
>
>> Hi List.
>>
>> Based on documentation and what I'm actually seeing now , system vms
>>can
>> only failover to other hosts in the same cluster and therefore pod.
>>
>> I'm guessing this is because the system vm is using a management ip
>> assigned to that specific pod.
>>
>> Is there any HA offering that could make system VMs to failover to
>> different pods/clusters?
>>
>> Thanks!
>>
Re: System VMs Failover
Posted by Kelven Yang <ke...@citrix.com>.
System VMs are tend to be running in stateless mode, instead of fail-over
of same VM, we are launching new instance of them, this behave is done for
storage and console proxy system VM, for virtual router system VM, there
is a redundant configuration, automatic HA would make the redundant
configuration unstable. so in general, system VMs don't use HA offering
purposely.
Kelven
On 4/11/13 7:54 AM, "Jeronimo Garcia" <ga...@gmail.com> wrote:
>Anyone ever bumped into this yet?
>
>Thanks!
>
>
>On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia
><ga...@gmail.com>wrote:
>
>> Hi List.
>>
>> Based on documentation and what I'm actually seeing now , system vms
>>can
>> only failover to other hosts in the same cluster and therefore pod.
>>
>> I'm guessing this is because the system vm is using a management ip
>> assigned to that specific pod.
>>
>> Is there any HA offering that could make system VMs to failover to
>> different pods/clusters?
>>
>> Thanks!
>>
Re: System VMs Failover
Posted by Jeronimo Garcia <ga...@gmail.com>.
Anyone ever bumped into this yet?
Thanks!
On Wed, Apr 10, 2013 at 11:27 AM, Jeronimo Garcia <ga...@gmail.com>wrote:
> Hi List.
>
> Based on documentation and what I'm actually seeing now , system vms can
> only failover to other hosts in the same cluster and therefore pod.
>
> I'm guessing this is because the system vm is using a management ip
> assigned to that specific pod.
>
> Is there any HA offering that could make system VMs to failover to
> different pods/clusters?
>
> Thanks!
>