Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1421548

Summary:	Failed host installs not handled in a user friendly way.
Product:	Red Hat Enterprise Virtualization Manager	Reporter:	Germano Veit Michel <gveitmic>
Component:	ovirt-engine	Assignee:	Martin Perina <mperina>
Status:	CLOSED ERRATA	QA Contact:	Petr Matyáš <pmatyas>
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	4.1.0	CC:	gklein, gveitmic, lsurette, lsvaty, mperina, rbalakri, Rhev-m-bugs, srevivo, ykaul
Target Milestone:	ovirt-4.1.1
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	Infra	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Germano Veit Michel 2017-02-13 04:08:22 UTC

Description of problem:

It's a bit annoying to Re-Install Hosts that failed installation in 4.1. I remember this being nicer in 4.0/3.6

Version-Release number of selected component (if applicable):
ovirt-engine-4.1.0.4-0.1.el7.noarch

How reproducible:
100%

Steps to Reproduce:
1. Add Host Through the GUI
2. Installation Fails
   * any reason
   * to reproduce please forget to subscribe the required channels, it will fail to find vdsm package

Now comes the annoying parts, see the comments:

3. Right click and select Re-Install, this message is displayed:
   -> Cannot edit Host. Host parameters cannot be modified while Host is operational.
   -> Please switch Host to Maintenance mode first.
   Comment: No, host is not operational, the install failed and I never activated it.
   I should be able to do this without moving to maintenance.

4. OK, I accept and go ahead and switch to maintenance. Click Maintenance.
   Comment: Always fails. It goes to Non-Responsive (100% of the time) 

5. Try Maintenance again.
   Comment: Now it succeeds (100% of the time)

6. Re-Install works

Actual results:
It's a bit cumbersome when install fails, should be easier for the admin to repeat the operation.

Expected results:
-> Should allow to Re-Install right away after failure.
-> That switch to maintenance mode should have succeeded (100% of the time I need to try twice)

Comment 2 Martin Perina 2017-02-13 10:53:13 UTC

Germano, are you sure that you tried to execute Reinstall when host is in status 'Install Failed'? I've tried that today with latest 4.1 and you can execute it without any issues (no need to switch host to Maintenance).

Moving host to Maintenance from InstallFailed (especially for new installations when host is not yet properly configured) is a bit problematic:

1. Unfortunately we execute HostMonitoring for status PreparingForMaintenance
2. When default firewall is present on the host, all packets for engine <-> VDSM communication are dropped, which cause NoRouteToHost exception and this moves host from PreparingToMaintenance to NonResponsive (we need to rethink if host monitoring shouldn't be disable for PreparingToMaintenance status)

Good news is that moving host to Maintenance from NonResponsive is almost immediate, so it's a workaround for that.

But as I said above, you should be able to execute Reinstall for host in InstallFailed status ...

Comment 3 Germano Veit Michel 2017-02-13 23:45:12 UTC

(In reply to Martin Perina from comment #2)
> Germano, are you sure that you tried to execute Reinstall when host is in
> status 'Install Failed'? I've tried that today with latest 4.1 and you can
> execute it without any issues (no need to switch host to Maintenance).

Yes, quite sure. I can reproduce it as many times I want. And it happens on different installs, including upstream oVirt. Always the same.

Just to illustrate I reproduced it again today and took a screenshot, including the events tab showing the failed install and the host being in Install Failed state. Then I clicked on Re-Install, it asked for the SSH password and I clicked OK. Same story.

Right, screenshots are not really useful. But what sort of information should I collect to help you troubleshoot this? Any extra debug level in the engine?

Comment 5 Martin Perina 2017-02-15 14:37:39 UTC

Neither me nor QA were able to reproduce this issue on latest 4.1 builds (ovirt-engine-4.1.1), so this was already fixed in patches merged after RHV 4.1 beta. So moving to MODIFIED to verify that properly

Comment 8 Petr Matyáš 2017-02-20 17:27:39 UTC

Verified on 4.1.1-2