Note: This is a beta release of Red Hat Bugzilla 5.0. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Also email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback here.
Bug 1510189 - LVM manual procedure for configuring a mirrored volume in a cluster should be redone for RAID
Summary: LVM manual procedure for configuring a mirrored volume in a cluster should be...
Alias: None
Product: Red Hat Enterprise Linux 7
Classification: Red Hat
Component: doc-Logical_Volume_Manager
Version: 7.4
Hardware: Unspecified
OS: Unspecified
Target Milestone: rc
: ---
Assignee: Steven J. Levine
QA Contact: Corey Marthaler
Depends On: 1507973
TreeView+ depends on / blocked
Reported: 2017-11-06 21:42 UTC by Steven J. Levine
Modified: 2019-03-06 00:44 UTC (History)
8 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of: 1507973
Last Closed:
Target Upstream Version:

Attachments (Terms of Use)

Description Steven J. Levine 2017-11-06 21:42:29 UTC
The history of this BZ is noted below.  The simple conclusion is that we need to redo section 5.5 in the LVM manual so that it shows a RAID volume in the example rather than an old-style mirror.  Also: Does this example need to show a clustered example?  It seems so, since there are some specific options people need to know about that are not part of the other cluster examples we have.  Needs to be pursued.

+++ This bug was initially created as a clone of Bug #1507973 +++

The LVM manual currently contains a procedure for configuring LVM in a cluster. That procedure assumes CLVM (with GFS2).  But this ignores the more common case of halvm.

The procedures for both LVM in a cluster and halvm are documented elsewhere (halvm in the HA Admin Guide, clvm in the gfs2 manual), so it might be better for the LVM manual just to summarize the differences but refer elsewhere for the actual procedures to avoid confusion.

--- Additional comment from Steven J. Levine on 2017-11-02 16:56:33 EDT ---

We already have a basic explanation of halvm vs clvm here:

This needs to be expanded and clarified, with references to the procedures elsewhere -- and we can drop the procedure we refer to now in this manual.

We have lots of info about halvm vs. clvm in the RHEL 6 cluster manual, which we can probably pillage:

But oddly we provide the procedure in two places in the current LVM manual for clustered LVM:


--- Additional comment from Steven J. Levine on 2017-11-03 17:26:11 EDT ---

Ok, I've completely rewritten the section in the LVM manual that used to be about CLVM and is now called "LVM Volumes in a Red Hat High Availability Cluster".  The draft is here:

I also reference this new section in the subsection in the CLI command chapter about creating volumes in a cluster.

But this is a MANY-part update and I now have to do the following:

- Update the procedure in the GFS2 manual that this references to include the installation step.

- I removed the CLVMD illustration as part of this rewrite, since I'm not sure how important it is anyway in the current procedure where Pacemaker starts CLVMD in the first place.  BUT I need to update the GFS2 procedure to mention that Pacemaker starts this and why it matters.

The removed illustration can still be seen here, as figure 1.2

- I will remove the section in chapter 3 on configuring logical volumes in a cluster, since we now reference the procedure in the cluster admin guide and this is really a cluster procedure rather than an LVM procedure per se.

- There is a procedure in the LVM manual on configuring mirrored volumes in a cluster which is only there because we had that procedure in the older versions of the LVM manual to show one example of clvmd so I needed to update it, but we have now added a better configuration example for CLVM in the cluster admin guide.  AND this procedure uses old-style mirroring which is still supported but we now use RAID mirroring so I should completely redo this section and turn it into a basic section on configuring an LVM RAID volume.

That should address all the points of the BZ here.  When I have finished all of that I will send the new section for review (and also ask if it was ok to have removed the illustration about clvmd).

--- Additional comment from Steven J. Levine on 2017-11-06 15:58:20 EST ---

I have called out the installation step (installing clvm and gfs2 packages) as step 1 in the procedure in the draft of the GFS2 manual, for clarity's sake. Previously this was summarized in the intro sentence (After installing the cluster software and GFS2 and LVM packages...)

I'll publish this as part of publishing the other updates in this BZ.

--- Additional comment from Steven J. Levine on 2017-11-06 16:34:40 EST ---

I have removed what had been section 3.1 in the LVM manual, creating LVM volumes in a cluster, since this procedure is also in the GFS2 manual.  A similar procedure will also be in the Cluster Admin guide when we publish the new chapter on configuring clustered Samba, but it seems to belong in the cluster doc rather than the LVM doc per se, and GFS2 is part of that.  This is the section that has been removed in the current draft:

But -- I have added some information to the GFS2 procedure that had been in the LVM manual so as not to lose it: At the end of step 7 there is some info about a volume in a cluster being visible to only one node in the cluster, and a warning about asymmetric cluster configurations

--- Additional comment from Steven J. Levine on 2017-11-06 16:39:03 EST ---

I am going to clone off a separate BZ to deal with the issue of rewriting the procedure for configuring a mirrored volume in a cluster, since that's a huge task, to rewrite this for RAID volumes, and will require some discussion with both LVM and cluster developers.  What's currently in the manual is correct so I'm leaving it for now, but the plans are to update that section.

Comment 3 Steven J. Levine 2017-12-12 21:58:01 UTC
This BZ is possibly a duplicate of BZ#1513721, but there seems to be some different information so I'm leaving them both open for now.

Note You need to log in before you can comment on or make changes to this bug.