Bug 711880

Summary:	unable to change read/write permissions of active cmirrors
Product:	Red Hat Enterprise Linux 5	Reporter:	Corey Marthaler <cmarthal>
Component:	cmirror	Assignee:	Jonathan Earl Brassow <jbrassow>
Status:	CLOSED WONTFIX	QA Contact:	Cluster QE <mspqa-list>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	5.7	CC:	agk, ccaulfie, dwysocha, edamato, heinzm, mbroz
Target Milestone:	rc
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-11-22 22:19:36 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 Corey Marthaler 2011-06-08 20:50:13 UTC

Description of problem:
This does however work for single machine mirrors.

# SINGLE MACHINE

[root@taft-03 ~]# lvcreate -m 1 -n mirror -L 100M single
  Logical volume "mirror" created
[root@taft-03 ~]# lvs -a -o +devices
  LV                VG       Attr   LSize   Log         Copy%  Devices
  mirror            single   mwi-a- 100.00M mirror_mlog 100.00 mirror_mimage_0(0),mirror_mimage_1(0)
  [mirror_mimage_0] single   iwi-ao 100.00M                    /dev/sdb1(0)
  [mirror_mimage_1] single   iwi-ao 100.00M                    /dev/sdc1(0)
  [mirror_mlog]     single   lwi-ao   4.00M                    /dev/sdh1(0)
[root@taft-03 ~]# lvchange -pr single/mirror
  Logical volume "mirror" changed
[root@taft-03 ~]# lvs -a -o +devices
  LV                VG       Attr   LSize   Log         Copy%  Devices
  mirror            single   mri-a- 100.00M mirror_mlog 100.00 mirror_mimage_0(0),mirror_mimage_1(0)
  [mirror_mimage_0] single   iwi-ao 100.00M                    /dev/sdb1(0)
  [mirror_mimage_1] single   iwi-ao 100.00M                    /dev/sdc1(0)
  [mirror_mlog]     single   lwi-ao   4.00M                    /dev/sdh1(0)


# CLUSTER

[root@grant-03 ~]# lvcreate -m 1 -n mirror -L 100M cluster
  Logical volume "mirror" created
[root@grant-03 ~]# lvs -a -o +devices
  LV                VG       Attr   LSize   Log         Copy%  Devices
  mirror            cluster  mwi-a- 100.00M mirror_mlog  68.00 mirror_mimage_0(0),mirror_mimage_1(0)
  [mirror_mimage_0] cluster  Iwi-ao 100.00M                    /dev/sdb1(0)
  [mirror_mimage_1] cluster  Iwi-ao 100.00M                    /dev/sdb2(0)
  [mirror_mlog]     cluster  lwi-ao   4.00M                    /dev/sdc3(0)
[root@grant-03 ~]# lvchange -pr cluster/mirror
  Cannot change permissions of mirror "mirror" while active.
[root@grant-03 ~]# lvs -a -o +devices
  LV                VG       Attr   LSize   Log         Copy%  Devices
  mirror            cluster  mwi-a- 100.00M mirror_mlog 100.00 mirror_mimage_0(0),mirror_mimage_1(0)
  [mirror_mimage_0] cluster  iwi-ao 100.00M                    /dev/sdb1(0)
  [mirror_mimage_1] cluster  iwi-ao 100.00M                    /dev/sdb2(0)
  [mirror_mlog]     cluster  lwi-ao   4.00M                    /dev/sdc3(0)


Version-Release number of selected component (if applicable):
2.6.18-265.el5

lvm2-2.02.84-4.el5    BUILT: Thu Jun  2 05:04:32 CDT 2011
lvm2-cluster-2.02.84-3.el5    BUILT: Wed Apr 27 03:42:43 CDT 2011
device-mapper-1.02.63-3.el5    BUILT: Thu May 19 08:09:22 CDT 2011
device-mapper-event-1.02.63-3.el5    BUILT: Thu May 19 08:09:22 CDT 2011
cmirror-1.1.39-10.el5    BUILT: Wed Sep  8 16:32:05 CDT 2010
kmod-cmirror-0.1.22-3.el5    BUILT: Tue Dec 22 13:39:47 CST 2009


How reproducible:
Everytime

Comment 2 Jonathan Earl Brassow 2011-11-22 22:19:36 UTC

I'm sorry, I'm backing out of this one...

The code that enforces the inability to change permissions is found in lvchange.c:
	if ((lv->status & MIRRORED) && (vg_is_clustered(lv->vg)) &&
	    lv_info(cmd, lv, 0, &info, 0, 0) && info.exists) {
		log_error("Cannot change permissions of mirror \"%s\" "
			  "while active.", lv->name);
		return 0;
	}
I'm sure you could simply remove this check and the permissions would be allowed to change.  However, I'm not clear what the goal is here and how the system is suppose to behave when there are multiple machines with file systems mounted - perhaps already with RW permissions.

Here is how things look in the single machine case:
[root@bp-01 ~]# mount /dev/vg/lv /mnt
[root@bp-01 ~]# lvchange -pr vg/lv
  Logical volume "lv" changed
[root@bp-01 ~]# touch /mnt/foo
[root@bp-01 ~]# ls -l /mnt
total 13
-rw-r--r-- 1 root root     0 Nov 22 16:09 foo
drwx------ 2 root root 12288 Nov 22 16:09 lost+found
[root@bp-01 ~]# umount /mnt
[root@bp-01 ~]# !mount
mount /dev/vg/lv /mnt
mount: block device /dev/vg/lv is write-protected, mounting read-only

So, in the single machine case, we are allowed to change the permissions while the device is open, but the file system can still proceed to use the block device using the permissions it was originally opened with.  It is only forced into read-only mode if a mount is attempted after the permission change.

It seems to me that it is the right thing to do to force users to deactivate the logical volume in order to change the permissions.  This forces them to unmount their file systems.  After all, what is the point in changing the permissions on their LV if they are allowed to keep running in the old mode?

Unless someone weighs in on this to convince me otherwise, I am closing this bug WONTFIX.  (Additionally, I would think it proper to open a new bug to address the exposed issue on single machine illustrated above, but that may actually be expected behavior...)