Bug 1219620 - upgrade to oVirt 3.5.2 breaks power management (at least idrac7)
Summary: upgrade to oVirt 3.5.2 breaks power management (at least idrac7)
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: oVirt
Classification: Retired
Component: ovirt-engine-core
Version: 3.5
Hardware: Unspecified
OS: Unspecified
unspecified
urgent
Target Milestone: ---
: 3.5.4
Assignee: Eli Mesika
QA Contact: movciari
URL:
Whiteboard: infra
Depends On:
Blocks: 1221312
TreeView+ depends on / blocked
 
Reported: 2015-05-07 19:53 UTC by jas
Modified: 2016-02-10 19:33 UTC (History)
10 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2015-09-03 13:54:54 UTC
oVirt Team: Infra
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
oVirt gerrit 41482 0 None None None Never

Internal Links: 1221316 1359875

Description jas 2015-05-07 19:53:31 UTC
Description of problem:

After upgrading from engine running CentOS 7 and ovirt node 3.5.1 to engine running on CentOS 7.1 and CentOS 7.1 hosts as "nodes", fencing with idrac7 no longer works.

Steps to Reproduce:
1. Add 2 new hosts to engine configured with idrac7 power management (no longer need to specify options for idrac7 with 3.5.2 - thank you.).
2. Click "Test" button on Power Management page.
3. "Test Succeeded, unknown".

It didn't really succeed at all.

engine:

2015-05-07 15:19:37,585 INFO  [org.ovirt.engine.core.bll.FenceExecutor] (ajp--127.0.0.1-8702-1) Using Host virt1 from cluster EECS-Primary as proxy to execute Status command on Host 
2015-05-07 15:19:37,587 INFO  [org.ovirt.engine.core.bll.FenceExecutor] (ajp--127.0.0.1-8702-1) Executing <Status> Power Management command, Proxy Host:virt1, Agent:ipmilan, Target Host:, Management IP:virt2-idrac, User:root, Options:privlvl=OPERATOR,lanplus,delay=10, Fencing policy:null
2015-05-07 15:19:37,602 INFO  [org.ovirt.engine.core.vdsbroker.vdsbroker.FenceVdsVDSCommand] (ajp--127.0.0.1-8702-1) START, FenceVdsVDSCommand(HostName = virt1, HostId = 6a638c0a-0a48-4e77-81b5-f89ae572cf3e, targetVdsId = 8df66929-0cb7-4347-b004-c5531178d024, action = Status, ip = virt2-idrac, port = , type = ipmilan, user = root, password = ******, options = 'privlvl=OPERATOR,lanplus,delay=10', policy = 'null'), log id: 682d66b2
2015-05-07 15:19:53,876 WARN  [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (ajp--127.0.0.1-8702-1) Correlation ID: null, Call Stack: null, Custom Event ID: -1, Message: Power Management test failed for Host virt2.Done
2015-05-07 15:19:53,876 INFO  [org.ovirt.engine.core.vdsbroker.vdsbroker.FenceVdsVDSCommand] (ajp--127.0.0.1-8702-1) FINISH, FenceVdsVDSCommand, return: Test Succeeded, unknown, log id: 682d66b2

on virt1 vdsm log:

hread-11121::DEBUG::2015-05-07 15:19:53,864::API::1164::vds::(fence) rc 1 inp agent=fence_ipmilan
ipaddr=virt2-idrac
login=root
action=status
passwd=XXXX
privlvl=OPERATOR
delay=10
lanplus out [] err ['Failed: Unable to obtain correct plug status or plug is not available', '', '']
Thread-11121::DEBUG::2015-05-07 15:19:53,864::API::1235::vds::(fenceNode) rc 1 in agent=fence_ipmilan
ipaddr=virt2-idrac
login=root
action=status
passwd=XXXX
privlvl=OPERATOR
delay=10
lanplus out [] err ['Failed: Unable to obtain correct plug status or plug is not available', '', '']

We can see that engine was able to contact virt1.
virt1 was able to contact virt2-idrac.
Password is correct.

I know that fence_ipmilan works successfully because I can go to virt2 and run the following:

/usr/sbin/fence_ipmilan -a virt1-idrac -l root -o status -p 'my-password' -L OPERATOR -P

The result is "Status: ON" as it was with 3.5.1.

Likewise, I can make virt1 verify virt2-idrac manually:

/usr/sbin/fence_ipmilan -a virt2-idrac -l root -o status -p 'my-password' -L OPERATOR -P 

Again, "Status: ON".

I've seen this "Test Succeeded, unknown" previously while getting the options right for idrac7 with 3.5.1.  The error needs to be corrected as well.  In this case, it was a failure.

Comment 2 Eli Mesika 2015-05-10 08:40:52 UTC
Marek, is there any change in CentOS 7.1 fence-agents RPM that requires the lanplus parameter to be defined with lanplus=1 or lanplus=0 rather than the  CentOS 7  fence-agents RPM that works OK with only lanplus 

Please reply to Comment 1

Comment 3 Marek Grac 2015-05-11 07:42:56 UTC
@Eli:

yes, it is very likely. Situation with 'lanplus' instead of 'lanplus=X' was not intended. Each option should have a value.

Comment 4 Oved Ourfali 2015-05-11 10:07:44 UTC
Marek - you must understand this causes a lot of regressions for us, and it wasn't communicated to us.
Please consider supporting both, to support old usages.
I'll open a bug on 7.1.

Comment 5 Marek Grac 2015-05-11 10:20:40 UTC
@Oved:

I understand but this was never meant to be used, it worked only as a side effect and only in this case&fence agent afaik. Are you using 'ssl' or 'secure' in other agents with =1 or without?

Comment 6 Oved Ourfali 2015-05-11 10:22:00 UTC
(In reply to Marek Grac from comment #5)
> @Oved:
> 
> I understand but this was never meant to be used, it worked only as a side
> effect and only in this case&fence agent afaik. Are you using 'ssl' or
> 'secure' in other agents with =1 or without?


I understand it wasn't the original intent, but it was already supported and done.

As for your question, Eli?

Comment 7 Eli Mesika 2015-05-11 10:56:21 UTC
(In reply to Oved Ourfali from comment #6)

> As for your question, Eli?

ssl & secure are passed without a value . i.e. ssl, secure not ssl=x secure=y 

is that also a problem ?

Comment 9 Eli Mesika 2015-05-31 09:01:01 UTC
putting comment 1 to private since it has a plain text password by mistake

Comment 10 Sandro Bonazzola 2015-07-02 06:51:28 UTC
Since oVirt 3.5.4 RC1 has been released, please ensure that the fix is included in the build and move the bug to ON_QA accordingly.

Comment 11 Sandro Bonazzola 2015-09-03 13:54:54 UTC
This is an automated message.
oVirt 3.5.4 has been released on September 3rd 2015 and should include the fix for this BZ. Moving to closed current release.


Note You need to log in before you can comment on or make changes to this bug.