[dm-devel] device mapper configurations

[Brian De Wolf]

Domenico Viggiani wrote:
>> -----Original Message-----
>> From: dm-devel-bounces at redhat.com 
>> [mailto:dm-devel-bounces at redhat.com] On Behalf Of Jimmy Stewpot
>> Sent: Wednesday, January 02, 2008 3:21 PM
>> ...
>> When the system boots I get loads of errors that look like this
>>
>> [  217.140126] end_request: I/O error, dev sdb, sector 0 [  
>> ...
>>
>> When the servers boot up. The items being reported are the 
>> underlying scsi devices which we have setup above. Is it 
>> normal to get these messages or do I need to explicitly place 
>> the underlying devices into the blacklist?
> 
> You have an active-passive array (like my EMC Clariion) and these messages
> are normal even if annoying. It's not much "professional".
> With Clariion, as far as I know, the only solution is upgrading Flare to
> ver. 26 that simulates active-active behaviour.

They should be expected, since you can't read the passive link, but it's only
failing on sector 0 which sounds like udev is trying to read the partition
table.  I think I remember fixing this behavior on a test box (that's long gone,
now, so I can't look it up) by fiddling with the udev rules to not bother
scanning the multipathed (i.e. fiber channel) devices.

Sorry for the lack of specifics, but it's been a while, and I wasn't very
specific in my own notes about what I did.  They should be preventable, though,
if I remember right.