Bug 618303

Summary: change default plock_ownership to 0
Product: Red Hat Enterprise Linux 6 Reporter: David Teigland <teigland>
Component: clusterAssignee: David Teigland <teigland>
Status: CLOSED CURRENTRELEASE QA Contact: Cluster QE <mspqa-list>
Severity: medium Docs Contact:
Priority: low    
Version: 6.0CC: ccaulfie, cluster-maint, djansa, fdinitto, lhh, rpeterso, ssaha, teigland
Target Milestone: rcKeywords: RHELNAK
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: cluster-3.0.12-19.el6 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2010-11-10 20:00:21 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:
Attachments:
Description Flags
the patch none

Description David Teigland 2010-07-26 15:40:17 UTC
Description of problem:

plock_ownership is an alternative, more complex, plock protocol in
dlm_controld that optimizes for a certain pattern of locking by
applications.

In RHEL5, plock_ownership defaults to 0 (off), and in RHEL6 it defaults to
1 (on).  on is more complicated and QE is still finding problems, i.e. off
is going to result in less customer problems.

Assuming both modes work correctly, neither is always a better choice.

on optimizes the case of repetitious locking of the same mostly unshared
(across nodes) files, at the expense of complexity and extra messages when
nodes are locking the same files.

off optimizes for the case of nodes generally acquiring locks on the same
files, at the expense of extra messages when nodes repeatedly lock their
own files that other nodes don't access.

At the beginning of RHEL6 development, we switched the default to on,
assuming, I suppose, that the new optimization would be an improvement
for most people, but I don't think that has been the case.

plock_ownership is one of the settings that cannot be changed online; the cluster needs to be stopped to change it.  This means it is not as quick and simple for people to change if they have problems with it, and it also means we cannot change the default setting in an update release.

Regarding the testing of the two modes, all (or most) RHEL5 testing has been done with it off, and this is substantially more testing than RHEL6 testing thus far with it on.  (The plock code is the same in RHEL5 and RHEL6, it's just the default protocol has changed.)

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


How reproducible:


Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info:

Comment 2 RHEL Program Management 2010-07-26 15:57:35 UTC
This issue has been proposed when we are only considering blocker
issues in the current Red Hat Enterprise Linux release.

** If you would still like this issue considered for the current
release, ask your support representative to file as a blocker on
your behalf. Otherwise ask that it be considered for the next
Red Hat Enterprise Linux release. **

Comment 4 David Teigland 2010-07-27 18:48:24 UTC
Created attachment 434798 [details]
the patch

Comment 7 Dean Jansa 2010-08-24 21:14:50 UTC
Man pages contain the new defaults.

Cluster defaults to off:

[root@marathon-01 cluster]# dlm_tool dump | grep plock_ownership

Comment 8 releng-rhel@redhat.com 2010-11-10 20:00:21 UTC
Red Hat Enterprise Linux 6.0 is now available and should resolve
the problem described in this bug report. This report is therefore being closed
with a resolution of CURRENTRELEASE. You may reopen this bug report if the
solution does not work for you.