Bug 1903052

Summary:	ansible-runner-service.cil selinux module is often needlessly re-installed, takes a long time
Product:	[oVirt] ovirt-engine	Reporter:	Yedidyah Bar David <didi>
Component:	Setup.EngineCommon	Assignee:	Martin Necas <mnecas>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Guilherme Santos <gdeolive>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	4.4.1	CC:	bugs, gdeolive, mperina
Target Milestone:	ovirt-4.4.5	Flags:	pm-rhel: ovirt-4.4+
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	ovirt-engine-4.4.5.5	Doc Type:	No Doc Update
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2021-03-22 10:22:28 UTC	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 Yedidyah Bar David 2020-12-01 08:33:28 UTC

Description of problem:

ansible-runner-service.cil is installed:

- by code in packaging/setup/plugins/ovirt-engine-common, so by every utility using common (including cleanup, rename, etc.)

- inside there, by code in base/ (so also if not on the engine machine, e.g. separate dwh machine)

- on every run, even if nothing changed (if selinux is enabled)

This is taking quite some time - 20-40 seconds, on the cases I noticed.

Instead, this should:

- Be moved to somewhere in setup/plugins/ovirt-engine-setup/ovirt-engine, assuming we need to run it only during engine-setup, and only on the engine machine. If unsure about the structure of this tree, check setup/plugins/README.

- Have a refined condition. At minimum, only if engine is enabled. Perhaps also check if the policy changed and run only then. Latter is more complex to do.

I noticed this in the past and largely ignored - only pushed [2] - and now open this bug because I noticed it again, while looking at what takes time during OST basic-suite, specifically 002_bootstrap (because I am working on porting the hosted-engine counterpart to pytest and consider re-copying from basic-suite instead of manually porting). In current last successful run [1], test_verify_engine_backup took 506 seconds. Among these, 184 seconds are taken just by installing this module, no less than 5 times! First by engine-cleanup, then 3 times by provisiondb (called by engine-backup --mode=restore), and then last one by engine-setup. Considering that we do this after already running engine-setup once in the beginning, all 5 of them are not needed. Even if we do not refine the condition in engine-setup, we can save 4.

[1] https://jenkins.ovirt.org/job/ovirt-system-tests_basic-suite-master_nightly/630/
[2] https://gerrit.ovirt.org/c/ovirt-engine/+/110060

Comment 1 Martin Necas 2021-02-02 17:07:35 UTC

Hi,
could you please check out the patch, I'm not sure how to reproduce the run with the engine-backup and with engine-cleanup.
I was able to get the info msg only from engine-setup.

Comment 2 Guilherme Santos 2021-02-23 16:26:04 UTC

Hi Yedidyah, could you pls confirm/share some reproduction steps and/or some notes for QE verification? As far as I could understand, we should validate that ansible-runner-service.cil is only installing during engine-setup (and not when the other cases you mentioned). Is that correct?
Thanks!

Comment 3 Yedidyah Bar David 2021-02-24 06:49:44 UTC

Hi!

(Please don't post private comments on oVirt bugs, thanks)

(In reply to Guilherme Santos from comment #2)
> Hi Yedidyah, could you pls confirm/share some reproduction steps and/or some
> notes for QE verification? As far as I could understand, we should validate
> that ansible-runner-service.cil is only installing during engine-setup (and
> not when the other cases you mentioned). Is that correct?

Yes, and also that it's not ran at all if you do not setup an engine (e.g.
if you set up dwh on a separate machine).

> Thanks!

YW :-)

Comment 5 Sandro Bonazzola 2021-03-22 12:57:22 UTC

This bugzilla is included in oVirt 4.4.5 release, published on March 18th 2021.

Since the problem described in this bug report should be resolved in oVirt 4.4.5 release, it has been closed with a resolution of CURRENT RELEASE.

If the solution does not work for you, please open a new bug report.