Bug 460809

Summary:	runlevel 3<->5 spurious service starts
Product:	[Fedora] Fedora	Reporter:	Need Real Name <lsof>
Component:	irqbalance	Assignee:	Neil Horman <nhorman>
Status:	CLOSED NOTABUG	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	rawhide	CC:	lsof
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2008-09-02 23:58:39 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Need Real Name 2008-09-01 11:37:24 UTC

Something odd is going on..

# init 5
starting portreserve
starting anacron
starting irqbalance

# init 3
starting portreserve
starting anacron
starting irqbalance

# init 5
starting portreserve
starting anacron
starting irqbalance

Comment 1 Bill Nottingham 2008-09-02 18:34:00 UTC

This is an issue with the scripts in question. Moving to one of the components, please file bugs separately against the others.

Comment 2 Neil Horman 2008-09-02 19:37:12 UTC

are you running this service on a system with either a single cpu, or a cpu with multiple cores that shares the same L2 cache?  If so, this is expected behavior. In those environments irqbalance shuts down automatically (after starting successfully) because it detects that balancing irq's is futile (as there is no way to improve cache affinity with only one cache).  As such in each run level, it detects that it is stopped , and restarts , only to stop again.

Irritating yet, but better than the alternatives (keeping a pid file around and not starting when you want it to, etc).

Comment 3 Need Real Name 2008-09-02 21:15:35 UTC

Yes I probably am (some Atom chip, MSI Wind U100).

Will move to one of the other packages for fixing then. Thanks for the info.

Comment 4 Neil Horman 2008-09-02 23:58:39 UTC

Yeah, thats it.  Sorry, wish I could make it more intuitive, but even if I disable the auto shutdown code, I get questions asking why irqbalance doesn't distribute irqs between cores, or some other useless operation.  IIRC, I've got a documentation update in the irqbalance man page slated for a future release that will call attention to this behavior explicitly.