Bug 722754

Summary:	[vdsm][error-handling][lvm-conf]vdsm should add disable_after_error_count in lvm.conf
Product:	Red Hat Enterprise Linux 6	Reporter:	Moran Goldboim <mgoldboi>
Component:	vdsm	Assignee:	Federico Simoncelli <fsimonce>
Status:	CLOSED ERRATA	QA Contact:	Moran Goldboim <mgoldboi>
Severity:	high	Docs Contact:
Priority:	unspecified
Version:	6.3	CC:	abaron, bazulay, danken, fsimonce, iheim, tdosek, ykaul
Target Milestone:	rc
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:	vdsm-4.9-87	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-12-06 07:31:45 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Moran Goldboim 2011-07-17 11:31:52 UTC

Description of problem:
in case of multipath path error lvm commands are taking a long time (in my deployment it took 4.5 minutes) adding disable_after_error_count parameter to lvm.conf should put the device as disabled after a specific amount of errors has happened and make lvm and vdsm respond quicker.

Version-Release number of selected component (if applicable):
vdsm-4.9-81.el6.x86_64

How reproducible:
always

Steps to Reproduce:
1.make a multipath path faulty
2.lvm pvs vgs takes around 4.5 -5 minutes (50 SDs connected)
3.
  
Actual results:


Expected results:


Additional info:
changing this config value on lvm.conf makes operations to run much faster.

Comment 2 Dan Kenigsberg 2011-07-17 11:59:22 UTC

How and when would the device be re-enabled?

Comment 3 Federico Simoncelli 2011-07-20 13:41:11 UTC

Some additional notes, I just talked with Moran and "multipath path error" should be intended as "the storage is unreachable through all its paths".
Looking at the disable_after_error_count code introduced with the patch:

http://sources.redhat.com/git/gitweb.cgi?p=lvm2.git;a=commitdiff;h=74b228ee945934c3b979cbb70a29b3a721f5c683

The error_count is a one-shot value for each lvm command.
Summarizing: using disable_after_error_count has no side effects (eg: it's not permanently disabling a device) and would improve the lvm responsiveness when one storage is completely unreachable.

Comment 4 Dan Kenigsberg 2011-07-21 07:00:15 UTC

which value should we put in disable_after_error_count so we do not have to many false negatives? Moran, this question is directly also to your team :-)

Comment 5 Federico Simoncelli 2011-07-21 07:55:30 UTC

BZ#722754 Limit lvm retries to broken devices
Change-Id: I74dfdea05943f72c7b89eba42246fc8f26bf0035

http://gerrit.usersys.redhat.com/730

Comment 7 Tomas Dosek 2011-08-10 06:12:16 UTC

Verified - vdsm-4.9-91 - disable_after_error_count parameter is now set to 3.

Comment 8 errata-xmlrpc 2011-12-06 07:31:45 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

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

http://rhn.redhat.com/errata/RHEA-2011-1782.html