On 04/13/2012 11:09 AM, Martin Hovmöller wrote:
On Thu, Apr 12, 2012 at 12:41 PM, Itamar Heim <iheim(a)redhat.com
<mailto:iheim@redhat.com>> wrote:
On 04/12/2012 01:01 PM, Martin Hovmöller wrote:
It looks to me like the options are dropped, especially since the
fenceNode line from the hypervisor says "options=". What's going
on here?
When I try to save the power management configuration, the
Options-field
is empty the next time I go to Edit Host.
Running fence_ipmilan manually works fine, both from the
hypervisors and
the manager.
sounds like a bug in the UI.
please check from api to rule out an engine bug.
btw, from this bug, it looks like it was only added to intended 3.1
release of ovirt in the ui?
Bug 782703 - [webadmin] No power management options in GUI
which was fixed mid-march
this also seem in same area (could be fixed by now, but you are not
using latest devel branch), so again, i'd check from api first:
Bug 784790 - [ovirt][engine][webadmin] - power management - slot
information is not being saved (bladecenter)
It does indeed seem to be an engine bug.
The cli gives me the following info:
oVirt shell (connected)]# show host rhevh1.domain
id : 4e6ccb2c-84c7-11e1-aecc-2c768aac9498
name : rhevh1.domain
address : 10.23.30.110
cluster-id : 99408929-82cf-4dc7-a532-9d998063fa95
cpu-name : AMD Opteron(tm) Processor 6174
cpu-speed : 2200.0
cpu-topology-cores : 6
cpu-topology-sockets : 1
iscsi-initiator : iqn.1994-05.com.redhat:a85815241216
ksm-enabled : True
memory : 25194135552
port : 54321
power_management-address : 10.0.102.150
power_management-enabled : True
power_management-options-option-name : power_wait
power_management-options-option-value: 4
power_management-options-option-name : lanplus
power_management-options-option-value: yes
power_management-type : ipmilan
power_management-username : root
status-state : up
storage_manager-priority : 5
storage_manager-valueOf : false
summary-active : 0
summary-migrating : 0
summary-total : 0
transparent_hugepages-enabled : True
type : rhel
So apparently the options are saved. They are lost if I edit the host
and don't manually add the options again before clicking Ok, so it seems
to be saving the options correctly, but not re-reading them. Manual
fencing by klicking Power Management -> Restart fails as well, even
though the options seem to be passwd correctly when I do that. The fact
that this option exists means that the first bug you pointed out doesn't
match, but the second one seems to be the same one I'm seeing.
2012-04-12 20:48:19,307 INFO
[org.ovirt.engine.core.bll.FencingExecutor] (http--0.0.0.0-8443-6)
Executing <Status> Power Management command, Proxy Host:rhevh2.domain,
Agent:ipmilan, Target Host:rhevh1.domain, Management IP:10.0.102.150,
User:root, Options:power_wait=4,lanplus=yes
2012-04-12 20:48:19,309 INFO
[org.ovirt.engine.core.vdsbroker.vdsbroker.FenceVdsVDSCommand]
(http--0.0.0.0-8443-6) START, FenceVdsVDSCommand(vdsId =
6fdbf336-84c8-11e1-82ef-2c768aac9498, targetVdsId =
4e6ccb2c-84c7-11e1-aecc-2c768aac9498, action = Status, ip =
10.0.102.150, port = , type = ipmilan, user = root, password = ******,
options = 'power_wait=4,lanplus=yes'), log id: 11a7cfb9
2012-04-12 20:48:21,529 INFO
[org.ovirt.engine.core.vdsbroker.vdsbroker.FenceVdsVDSCommand]
(http--0.0.0.0-8443-6) FINISH, FenceVdsVDSCommand, return: Test Failed,
Host Status is: unknown. The fence-agent script reported the following
error: Getting status of IPMI:10.0.102.150...Chassis power = Unknown
Failed
, log id: 11a7cfb9
2012-04-12 20:48:21,529 ERROR
[org.ovirt.engine.core.bll.RestartVdsCommand] (http--0.0.0.0-8443-6)
Failed to run RestartVdsCommand on vds :rhevh1.domain
2012-04-12 20:48:21,530 WARN
[org.ovirt.engine.core.bll.RestartVdsCommand] (http--0.0.0.0-8443-6)
CanDoAction of action RestartVds failed.
Reasons:VDS_FAILED_FENCE_VIA_PROXY_CONNECTION,VDS_FENCING_OPERATION_FAILED,VAR__TYPE__HOST,VAR__ACTION__RESTART
I'm trying to figure out how to try to fence a machine from the cli, but
I can't figure out the correct syntax. The help seems to be a bit broken:
[oVirt shell (connected)]# action host rhevh1.domain fence
error:
status: 400
reason: Bad Request
detail: Action [fenceType] required for fence
Why do I need to specify fenceType when it's already stored for the host
(and how do I specify it for that matter)?
from:
https://xxx/api/capabilities
<fence_types>
<fence_type>manual</fence_type>
<fence_type>restart</fence_type>
<fence_type>start</fence_type>
fence_type>stop</fence_type>
<fence_type>status</fence_type>
</fence_types>