Heads up: mdadm-3.0-0.devel2.1.fc11 added to f11

[Radek Vykydal]

Doug Ledford wrote:
> On Mon, 2009-02-16 at 11:39 +0100, Radek Vykydal wrote:
>   
>> Doug Ledford wrote:
>>     
>>> The upgrade from mdadm 2.6.7.1 to mdadm 3.0 is not an entirely automated
>>> process.  The mdadm 3.0 package includes new naming rules on md devices
>>> (specifically, only md? and md_d? devices are allowed in /dev, all
>>> non-standard named md devices must be moved to /dev/md/).  In addition,
>>> I used upstream's udev rules file.  It has been revamped to support some
>>> of the new features of mdadm 3.0 and is different from our previous udev
>>> rules file.
>>>   
>>>       
>> The udev rules file for md devices is in udev package:
>>
>> [root at dhcp92 module]# diff /etc/udev/rules.d/64-md-raid.rules 
>> /lib/udev/rules.d/64-md-raid.rules
>> 14c14,15
>> < ATTR{md/array_state}=="|clear|inactive", GOTO="md_end"
>> ---
>>  > ATTR{md/array_state}=="clear|inactive", GOTO="md_end"
>>  > ATTR{md/array_state}!="?*", GOTO="md_end"
>> 18c19
>> < ENV{DEVTYPE}=="disk", ENV{MD_NAME}=="?*", 
>> SYMLINK+="disk/by-id/md-name-$env{MD_NAME}"
>> ---
>>  > ENV{DEVTYPE}=="disk", ENV{MD_NAME}=="?*", 
>> SYMLINK+="disk/by-id/md-name-$env{MD_NAME}", 
>> OPTIONS+="string_escape=replace"
>> 21c22
>> < ENV{DEVTYPE}=="partition", ENV{MD_NAME}=="?*", 
>> SYMLINK+="disk/by-id/md-name-$env{MD_NAME}-part%n"
>> ---
>>  > ENV{DEVTYPE}=="partition", ENV{MD_NAME}=="?*", 
>> SYMLINK+="disk/by-id/md-name-$env{MD_NAME}-part%n", 
>> OPTIONS+="string_escape=replace"
>>
>> where it contains also recently added patch to handle remove event added
>> to md layer (the first hunk).
>> (http://article.gmane.org/gmane.linux.hotplug.devel/13666)
>> Isn't udev package the right place for the rules file (and the
>> other two hunks change)?
>>     
>
> In the past, the udev maintainer has told me that the reason for
> creating the rules.d directory was specifically to get various packages
> to maintain their own rules as opposed to them all being in the udev
> package.  From that standpoint, it belongs in mdadm.
>
>   
Ok, my concern is a fix to appear in 64-md-raid.rules we use in F11.
I sent the patch to Harald Hoyer, he applied the patch to fedora udev
package, udev upstream applied it (modified), and it will
appear in udev-138 package. So, taking the rules file from mdadm
package, we need to use mdadm upstream having rules file
from udev-138, or use patched rules file from udev package.

The patch fixes removing of sysfs subtree with udev after stopping array
and we need it for anaconda storage rewrite for F11.
(http://git.kernel.org/?p=linux/hotplug/udev.git;a=blobdiff;f=rules/packages/64-md-raid.rules;h=def184f42c224ef04a7f5eb32398c3561e5fa265;hp=6fe4d46dc9741838ded6de63dae940aba87d32e6;hb=b822542608326092e177fd1707ca7fb53b2846c4;hpb=a402404fb23195839a9e008dbbe5edb5349c05b0)