Bug 1044862

Summary:	On OpenShift Enterprise 2.0 installation, oo-diagnostics -v -w 1 complains about kernel.sem semaphores too low: 128 < 512
Product:	OpenShift Container Platform	Reporter:	Jan Pazdziora <jpazdziora>
Component:	Containers	Assignee:	Luke Meyer <lmeyer>
Status:	CLOSED DEFERRED	QA Contact:	libra bugs <libra-bugs>
Severity:	low	Docs Contact:
Priority:	low
Version:	2.0.0	CC:	bleanhar, jpazdziora, libra-onpremise-devel, lmeyer
Target Milestone:	---	Keywords:	Regression
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2014-03-21 19:33:34 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Jan Pazdziora 2013-12-19 06:47:04 UTC

Description of problem:

On a fresh all-on-one OpenShift Enterprise 2.0 installation, running oo-diagnostics -v -w 1 fails with

INFO: running oo-accept-node
[31mFAIL: run_script
oo-accept-node had errors:
--BEGIN OUTPUT--
FAIL: kernel.sem semaphores too low: 128 < 512
1 ERRORS

--END oo-accept-node OUTPUT--[0m

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

Enterprise 2.0.

How reproducible:

Deterministic.

Steps to Reproduce:
1. Run ./openshift.sh to configure all on one Enterprise 2.0.
2. Run oo-diagnostics -v -w 1

Actual results:

INFO: running oo-accept-node
[31mFAIL: run_script
oo-accept-node had errors:
--BEGIN OUTPUT--
FAIL: kernel.sem semaphores too low: 128 < 512
1 ERRORS

Expected results:

No error.

Additional info:

I should point out that I've reported this issue on Origin back in June in bug 974497, it even got cloned to Enterprise product as bug 1007752 but sadly that got deferred, so this bug confirms my concern from bug 1007752 comment 4 -- the issue got to GA'ed version.

Comment 4 Luke Meyer 2013-12-19 23:00:48 UTC

That didn't use to happen. I'll have to figure out what changed.

Comment 5 Luke Meyer 2013-12-20 21:00:05 UTC

This shouldn't happen at all. When you use openshift.sh, it sets the semaphores higher, and has for as long as we've had it I think. See:

configure_sysctl_on_node()
{
  set_sysctl kernel.sem '250  32000 32  4096' 'Accomodate many httpd instances for OpenShift gears.'


Now, this doesn't take effect until system reboot (or there's a sysctl option to put it into effect). However the instructions are to reboot after install. Are you running this after a reboot or before?

Comment 6 Jan Pazdziora 2013-12-22 09:53:10 UTC

(In reply to Luke Meyer from comment #5)
> This shouldn't happen at all. When you use openshift.sh, it sets the
> semaphores higher, and has for as long as we've had it I think. See:
> 
> configure_sysctl_on_node()
> {
>   set_sysctl kernel.sem '250  32000 32  4096' 'Accomodate many httpd
> instances for OpenShift gears.'
> 
> 
> Now, this doesn't take effect until system reboot (or there's a sysctl
> option to put it into effect). However the instructions are to reboot after
> install. Are you running this after a reboot or before?

Yeah, thank you for he catch, it was the "not rebooting" thing.

I cannot reasonably reboot after installation so that was the cause of the bug. Running sysctl -p fixes the problem. I wonder if the openshift.sh could do that automatically?

Comment 8 Brenton Leanhardt 2014-03-21 19:33:34 UTC

I'm going to close this as DEFERRED.  If it ever happens it will be part of this PR: https://github.com/openshift/openshift-extras/pull/311