Bug 1368766

Summary: [Docs][Install] Changing host type not possible when host in maintenance mode
Product: Red Hat Enterprise Virtualization Manager Reporter: Nelly Credi <ncredi>
Component: DocumentationAssignee: rhev-docs <rhev-docs>
Status: CLOSED WONTFIX QA Contact: rhev-docs <rhev-docs>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 4.0.0CC: adahms, bugs, gklein, lbopf, lsurette, mperina, rbalakri, srevivo, ykaul, ylavi
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Known Issue
Doc Text:
Cause: It's not supported to allow changing host type (aka reprovisioning the host for example from NGN/RHV to Centos/RHEL) when host is in Maintenance status. If users want to do such operation, they need to do following steps: 1. Put host to Maintenance and then Remove the host form engine 2. Reprovision the host 3. Add reprovisioned host to engine It would be very complicated in terms of code amount to fail on all possible scenarios, so we should mention in documentation that above operation is not supported and may cause unexpected behaviour/issue. Consequence: Workaround (if any): Result:
Story Points: ---
Clone Of: Environment:
Last Closed: 2016-11-01 13:12:48 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: Docs RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Attachments:
Description Flags
host deploy log
none
engine log none

Description Nelly Credi 2016-08-21 09:14:01 UTC
Created attachment 1192558 [details]
host deploy log

Description of problem:
The engine tries to install NGN packages on RHEL host
the host was NGN and was provisioned to RHEL
while it stayed connected to the engine (host remove was not done)

Version-Release number of selected component (if applicable):


How reproducible:
50%

Steps to Reproduce:
1. Have NGN host connected to an engine
2. reprovision host to RHEL
3. upgrade icon appears
4. press upgrade

Actual results:
the engine tries to install NGN packages:
redhat-virtualization-host-image-update.noarch 
imgbased.noarch 
the host moves to status 'install failed'
can be activated again and the upgrade icon will stay there

Expected results:
the engine should know the host is RHEL now


Additional info:
looking at vds_dynamic table the engine knows its a RHEL host
(and in the UI as well)

Comment 1 Nelly Credi 2016-08-21 09:18:49 UTC
Created attachment 1192559 [details]
engine log

it happened at around 11am on the 19th

Comment 2 Martin Perina 2016-08-22 04:49:45 UTC
If the host was installed as NGN in webadmin, it will stay NGN, even though you have tried to install normal RHEL packages on it. The type of the host is determined during host deploy and you can check it:

1. In db running checking column vds_type in vds_static table:
     0 - RHEL
     1 - NGN
     2 - Vintage Node

2. In RESTAPI this is reported in host details within HostType

You were looking into vds_dynamic table where we have two columns:

  host_os
    - this is name-version-release reported by VDSM, but here you will always see RHEL even when it's NGN (we want to report underlying RHEL version here)
  pretty_name
    - here full description of host OS, there should be NGN and its correct version

I'm not aware of any way host to change OS type from NGN to RHEL other than removing the host and re-adding it again.

Comment 3 Nelly Credi 2016-08-22 09:27:44 UTC
im not installing rhel packages on the host, im doing reprovision to rhel & reinstall in the manager, i would expect reinstall to update this info, just as the UI is updated (the software tab)

reinstall definitely allows to install a host with different os, so i can imagine 2 options:
1. reinstall will update this info as it updates the software info
2. reinstall shouldnt allow reinstall of different OS

what do you think?

Comment 4 Martin Perina 2016-08-24 10:23:22 UTC
Allowing host type change during reinstallation may cause side effects, so we will add checks to host deploy process which will fail reinstallation of a host if host type change is detected (for example switching from NGN to RHEL or from Fedora to Centos).

If someone wants to change host type, the only correct way how to do that is remove the host from engine, reinstall and add it to engine again.

Comment 5 Ravi Nori 2016-09-08 13:55:07 UTC
There are several paths that need to be fixed if we intent to support detecting vds type change. The user can put the host in maintenance, reprovision the host and then activate/upgrade/reinstall the host. Fixing all the paths needs changes to code in engine as well as in otopi.

In my opinion we should not allow the user to reprovision the host, only way to change the host type should be to remove the host, reinstall and add it to engine again.

Comment 6 Martin Perina 2016-09-09 07:31:46 UTC
Because it's quite a corner case and the amount of code needed to fail correctly on all possible reprovisioning scenarios would be very high, we will not fix this in our codebase. But we need to mention in our documentation that reprovisioning the host while it's in Maintenance with different OS is not supported (more info in Doc Text).

Moving bug to Doc Team

Comment 7 Lucy Bopf 2016-10-26 07:12:50 UTC
Updating product and assignee information to allow proper tracking.

Comment 8 Yaniv Lavi 2016-11-01 13:12:48 UTC
Customer will either use RHV-H or RHEL and not mix and match as we do in testing.
Therefore this is so unlikely to happen and I'm closing this request.