[ovirt-users] Info on host not responding event and related actions

[Piotr Kliczewski]

+Marin to answer fencing configuration settings

On Tue, May 23, 2017 at 3:03 PM, Gianluca Cecchi
<gianluca.cecchi at gmail.com> wrote:
> Hello,
> I have an oVirt 4.1.1 environment with:
>
> - engine is a vSphere CentOS 7.3 VM with its nic on say vlan1
> - 2 x hosts (CentOS 7.3) with their ovirtmgmt lan on a bonding
> (active-backup) on say vlan2
>
> network architecture layout is to put hypervisors and mgmt servers in
> different vlans
>
> Today we had these 4 events below shown in our engine, with root cause
> apparently a maintenance network routing activity (it should have been
> transparent, network guys told..., but this is another story ;-)
> No alert message inside VMs
>
> 4) May 23, 2017 1:43:58 PM Host ov300 power management was verified
> successfully.
> 3) May 23, 2017 1:43:58 PM Status of host ov300 was set to Up.
> 2) May 23, 2017 1:43:55 PM Executing power management status on Host ov300
> using Proxy Host ov301 and Fence Agent ipmilan:10.10.193.103.
> 1) May 23, 2017 1:43:37 PM Host ov300 is not responding. It will stay in
> Connecting state for a grace period of 61 seconds and after that an attempt
> to fence the host will be issued.

Can you check in the logs what caused the host to become not responding?

>
> Can anyone tell exactly the meaning of the different lines?
> Is the 1) detected because the engine, from only a network point of view,
> was not able to ping/reach the hostname of the host ov300, or the "not
> responding" is any particular specific check?
> Is the "61 seconds" delay tunable?
> Is 2) an additional check to verify status of ov300?
> In case of failure of test in 2) would the fencing have been immediate or
> the delay described in 1) would have taken place?
> Are 3) and 4) messages independent from the engine being able to reach ov300
> or the 61 seconds delay would have been true anyway?
>
> Hope I have explained my doubts related to events that could determine a
> potential fencing of an active node with its running VMs... with the "only"
> temporary problem of connectivity between the engine and one of the nodes...
>
> Thanks in advance,
> Gianluca
>
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>