All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 0 of 12] LVM add 'mirrored' log type
@ 2010-02-05 20:34 Jonathan Brassow
  2010-02-09  0:28 ` Takahiro Yasui
  2010-02-11 19:42 ` Takahiro Yasui
  0 siblings, 2 replies; 4+ messages in thread
From: Jonathan Brassow @ 2010-02-05 20:34 UTC (permalink / raw)
  To: lvm-devel

The following set of patches adds the 'mirrored' log type.  It differs
from the last batch by changing the name of the log type
(s/redundant/mirrored/), fixing all the issues uncovered by the
testsuite, adding new test cases, and including log monitoring (provided
by Malahal Naineni).

 brassow




^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH 0 of 12] LVM add 'mirrored' log type
  2010-02-05 20:34 [PATCH 0 of 12] LVM add 'mirrored' log type Jonathan Brassow
@ 2010-02-09  0:28 ` Takahiro Yasui
  2010-02-09  1:09   ` malahal
  2010-02-11 19:42 ` Takahiro Yasui
  1 sibling, 1 reply; 4+ messages in thread
From: Takahiro Yasui @ 2010-02-09  0:28 UTC (permalink / raw)
  To: lvm-devel

On 02/05/10 15:34, Jonathan Brassow wrote:
> The following set of patches adds the 'mirrored' log type.  It differs
> from the last batch by changing the name of the log type
> (s/redundant/mirrored/), fixing all the issues uncovered by the
> testsuite, adding new test cases, and including log monitoring (provided
> by Malahal Naineni).

I'm testing the patch set, but a log monitoring doesn't work properly.
When dmeventd detects an error, lvconvert is executed for its repair.
For example, in case that the mirror VG/LV name is vg00/lv00, lvconvert
was called with the following options:

lvconvert --config devices{ignore_suspended_devices=1} --repair \
          --use-policies vg00/lv00_mlog

However, lvconvert can't accept vg00/lv00_mlog because "_mlog" is
an reserved name. When lvconvert is executed by itself, the following
message is displayed, and no repair is done.

# lvconvert --repair vg00/lv00_mlog
  Names including "_mlog" are reserved. Please choose a different LV name.
  Run `lvconvert --help' for more information.

We need to escape from this restriction when "mirrored" log is used.

Thanks,
Taka



^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH 0 of 12] LVM add 'mirrored' log type
  2010-02-09  0:28 ` Takahiro Yasui
@ 2010-02-09  1:09   ` malahal
  0 siblings, 0 replies; 4+ messages in thread
From: malahal @ 2010-02-09  1:09 UTC (permalink / raw)
  To: lvm-devel

Takahiro Yasui [tyasui at redhat.com] wrote:
> On 02/05/10 15:34, Jonathan Brassow wrote:
> > The following set of patches adds the 'mirrored' log type.  It differs
> > from the last batch by changing the name of the log type
> > (s/redundant/mirrored/), fixing all the issues uncovered by the
> > testsuite, adding new test cases, and including log monitoring (provided
> > by Malahal Naineni).
> 
> I'm testing the patch set, but a log monitoring doesn't work properly.
> When dmeventd detects an error, lvconvert is executed for its repair.
> For example, in case that the mirror VG/LV name is vg00/lv00, lvconvert
> was called with the following options:
> 
> lvconvert --config devices{ignore_suspended_devices=1} --repair \
>           --use-policies vg00/lv00_mlog
> 
> However, lvconvert can't accept vg00/lv00_mlog because "_mlog" is
> an reserved name. When lvconvert is executed by itself, the following
> message is displayed, and no repair is done.
> 
> # lvconvert --repair vg00/lv00_mlog
>   Names including "_mlog" are reserved. Please choose a different LV name.
>   Run `lvconvert --help' for more information.
> 
> We need to escape from this restriction when "mirrored" log is used.

I used an older LVM where it failed to convert a normal mirror. I used
"vgreduce" and it worked fine. So I used "vgreduce" to test the 'mlog'
conversion too. The actual command I used is constructed from the
following code piece:

snprintf(cmd_str, CMD_SIZE, "vgreduce --removemissing --force %s", vg)

I will test with the latest LVM and fix it. Thank you.

--Malahal.



^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH 0 of 12] LVM add 'mirrored' log type
  2010-02-05 20:34 [PATCH 0 of 12] LVM add 'mirrored' log type Jonathan Brassow
  2010-02-09  0:28 ` Takahiro Yasui
@ 2010-02-11 19:42 ` Takahiro Yasui
  1 sibling, 0 replies; 4+ messages in thread
From: Takahiro Yasui @ 2010-02-11 19:42 UTC (permalink / raw)
  To: lvm-devel

Jonathan Brassow wrote:
> The following set of patches adds the 'mirrored' log type.  It differs
> from the last batch by changing the name of the log type
> (s/redundant/mirrored/), fixing all the issues uncovered by the
> testsuite, adding new test cases, and including log monitoring (provided
> by Malahal Naineni).

I found other two cases which caused Segmentation Fault.

# lvcreate --version
  LVM version:     2.02.61(1)-cvs (2010-02-02)
  Library version: 1.02.44-cvs (2010-02-02)
  Driver version:  4.16.0

Tests are done with lvm2, 2.02.61(1)-cvs (2010-02-02) with this
patch set, and vg00 contains just one PV.

[case 1]
# lvcreate -m1 -L12m -nlv00 --mirrorlog mirrored --alloc anywhere vg00
Segmentation fault

[case 2]
# dmsetup table
vg00-lv00: 0 24576 linear 8:32 384
# lvconvert -m1 --mirrorlog mirrored --alloc anywhere vg00/lv00
Segmentation fault

In both cases, initialization of log_areas array seems to have a problem.

Thanks,
Taka




^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2010-02-11 19:42 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-02-05 20:34 [PATCH 0 of 12] LVM add 'mirrored' log type Jonathan Brassow
2010-02-09  0:28 ` Takahiro Yasui
2010-02-09  1:09   ` malahal
2010-02-11 19:42 ` Takahiro Yasui

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.