[ovirt-users] Ovirt Node Heartbeat Settings

[Martin Perina]

Hi Punit,

if you are using oVirt 3.5, there are 3 new options which could help
hosts not to be fenced. In cluster detail dialog there's new tab
Fencing Policy with 3 new options:

  1. Enable Fencing
    - this option if unchecked disables fencing for all hosts in cluster
    - but this should be used with caution, it should be turned off only,
      if you are expecting some networking problems between engine and hosts
      (such as router problems, ...). Because if you disable fencing and
      host will be marked as Non Responsible, HA VMs won't be restarted on
      new host
    - this option (if unchecked) also disables further options

  2. Skip fencing if host has live lease on storage
    - this option if checked will skip fencing of host if host is connected
      to storage (to be precise if the host renews its storage lease in the
      last 90 seconds, by default each host should renew its storage lease
      every 60 seconds)

  3. Skip fencing if cluster networking in unstable
    - this options if checked will skip fencing of host if amount of hosts
      in state Connection or Non Responsive in cluster is higher than
      selected amount in percents

So I think option 2 is best solution for your issue.

Martin

----- Original Message -----
> From: "Punit Dambiwal" <hypunit at gmail.com>
> To: "Martin Perina" <mperina at redhat.com>
> Cc: users at ovirt.org, "Dan Kenigsberg" <danken at redhat.com>, "Itamar Heim" <iheim at redhat.com>
> Sent: Wednesday, December 17, 2014 3:05:25 AM
> Subject: Re: [ovirt-users] Ovirt Node Heartbeat Settings
> 
> Hi Martin,
> 
> Yes...sometime the engine through the message in the event tab that host
> not responding...but actually it's working fine....because of the engine
> consider it as not responding...fence it....as i am using compute and
> storage on the same node (glusterfs)...it cause me serious downtime in the
> production...
> 
> Thanks,
> Punit
> 
> On Tue, Dec 16, 2014 at 4:51 PM, Martin Perina <mperina at redhat.com> wrote:
> >
> >
> >
> > ----- Original Message -----
> > > From: "Punit Dambiwal" <hypunit at gmail.com>
> > > To: "Martin Perina" <mperina at redhat.com>
> > > Cc: users at ovirt.org, "Dan Kenigsberg" <danken at redhat.com>, "Itamar
> > Heim" <iheim at redhat.com>
> > > Sent: Tuesday, December 16, 2014 9:13:24 AM
> > > Subject: Re: [ovirt-users] Ovirt Node Heartbeat Settings
> > >
> > > Hi Martin,
> > >
> > > Yes...I want to know the timeout after the host will be considered non
> > > responsive and fencing will be executed on it ?? can it be modify ?? if
> > > yes...please suggest me know the way to do this ??
> >
> > Please take a look at calculation in
> >
> > http://www.ovirt.org/Automatic_Fencing#Automatic_Fencing
> >
> > Host status will to Non Responding if either number of unsuccessful
> > connection attempts exceeds VDSAttemptsToResetCount or timeout from
> > last successful connection is expired.
> >
> > To alter time for host to be fence you can change these values from above
> > mentioned calculation can be changed using engine-config tool:
> >
> >   VDSAttemptsToResetCount
> >   TimeoutToResetVdsInSeconds
> >   DelayResetPerVmInSeconds
> >   DelayResetForSpmInSeconds
> >
> >
> > Is there any specific issue for which you need to change those values?
> >
> > Martin
> >
> > >
> > > Thanks,
> > > Punit
> > >
> > > On Tue, Dec 16, 2014 at 2:50 PM, Martin Perina <mperina at redhat.com>
> > wrote:
> > > >
> > > > Hi,
> > > >
> > > > not sure if I understand correctly. Do you want to know the timeout
> > after
> > > > the host
> > > > will be considered non responsive and fencing will be executed on it?
> > If
> > > > so please
> > > > take a look at those:
> > > >
> > > > http://www.ovirt.org/Automatic_Fencing
> > > > http://www.ovirt.org/Fence_kdump#Fencing_flow_with_fence_kdump
> > > >
> > > > It's quite a complex part of engine so feel free to ask
> > > >
> > > > Thanks
> > > >
> > > > Martin
> > > >
> > > > ----- Original Message -----
> > > > > From: "Punit Dambiwal" <hypunit at gmail.com>
> > > > > To: users at ovirt.org, "Dan Kenigsberg" <danken at redhat.com>, "Itamar
> > > > Heim" <iheim at redhat.com>
> > > > > Sent: Monday, December 15, 2014 9:37:47 AM
> > > > > Subject: [ovirt-users] Ovirt Node Heartbeat Settings
> > > > >
> > > > > Hi,
> > > > >
> > > > > I want to know more about the Ovirt engine and host node heartbeat
> > > > settings
> > > > > and interval....that means how and when the engine node consider the
> > host
> > > > > node as dead or shoot the command to fence it for reboot....
> > > > >
> > > > > Is there any way to modify those interval and settings to prevent the
> > > > > false-positive etc...
> > > > >
> > > > > Thanks,
> > > > > Punit
> > > > >
> > > > > _______________________________________________
> > > > > Users mailing list
> > > > > Users at ovirt.org
> > > > > http://lists.ovirt.org/mailman/listinfo/users
> > > > >
> > > >
> > >
> >
>